Natio al I dustrial Competitive ess through E ergy, E viro me t a d Eco omics NICE 3 Gra t No. DE-FG48-96R8-10598

Adva ced Process A alysis for Source Reductio i the Sulfuric Acid Petroleum Refi i g Alkylatio Process

Fi al Tech ical Report

by

Michael K. Rich Motiva E terprises

David McGee Louisia a Departme t of Natural Resources

Jack R. Hopper a d Carl L. Yaws Lamar U iversity

Derya B. Özyurt a d Ralph W. Pike Louisia a State U iversity

A joi t project with Louisia a Departme t of Natural Resources, Motiva E terprises, a d Gulf Coast Hazardous Substa ce Research Ce ter

submitted to the

U. S. Departme t of E ergy

September 1, 2001

Gulf Coast Hazardous Substa ce Research Ce ter Lamar U iversity Beaumo t, Texas 77710

TABLE F C NTENTS

ABSTRACT v

LIST OF TABLES vi

LIST OF FIGURES vii

CHAPTER 1. INTRODUCTION 1 Overview of Adva ced Process A alysis System 1 Flowsheet Simulatio 3 O -Li e Optimizatio 4 Chemical Reactor A alysis 6 Pi ch A alysis 7 Pollutio Assessme t 9 Alkylatio 10 Summary 11

CHAPTER 2. LITERATURE REVIEW 12 Adva ced Process A alysis System 12 I dustrial Applicatio s of O -Li e Optimizatio 13 Key Eleme ts of O -Li e Optimizatio 14 E ergy Co servatio 15 Pollutio Preve tio 16 Sulfuric Acid Alkylatio Process 22 Alkylatio i Petroleum I dustry 23 Commercial Sulfuric Acid Alkylatio Process 24 Theory of Alkylatio Reactio s 26 Reactio Mecha ism for Alkylatio of Isobuta e with Propyle e 30 Reactio Mecha ism for Alkylatio of Isobuta e with Butyle e a d Pe tyle e 30 I flue ce of Process Variables 34 Feedstock 36 Products 36 Catalysts 37 Summary 38

CHAPTER 3. METHODOLOGY 39 Flowsheeti g Program 40 O -Li e Optimizatio Program 43 Chemical Reactor A alysis Program 47 Heat Excha ger Network Program 48 Pollutio I dex Program 51 Summary 53

ii

CHAPTER 4. MOTIVA ALKYLATION PROCESS 54 Process Descriptio 54 STRATCO Co tactor 55 Refrigeratio Sectio 59 Depropa izer 61 Alkylate Deisobuta izer 61 Saturate Deisobuta izer 62 Process Simulatio 62 STRATCO Co tactor (5C-623) 66 Acid Settler (5C-631) 72 Depropa izer (5C-603) 73 Alkylate Deisobuta izer (5C-606) 76 Suctio Trap/Flash Drum (5C-614) 79 Eco omizer (5C-616) 82 Compressor (5K-601) 82 Olefi Feed-Efflue t Excha ger (5E-628) 85 Model Validatio 87 Summary 88

CHAPTER 5. RESULTS 89 Flowsheet Simulatio 89 Process Flow Diagram 89 Co strai t Equatio s 90 Measured Variables 91 U measured Variables 92 Parameters 92 Co sta ts 93 E thalpy Tables 93 Flowsheet Simulatio Summary 94 O -Li e Optimizatio 94 Gross Error Detectio a d Data Validatio 94 Parameter Estimatio a d Data Reco ciliatio 98 Eco omic Optimizatio 100 Heat Excha ger Network Optimizatio 103 Pollutio Assessme t 109 Summary 111

CHAPTER 6 CONCLUSIONS 112

REFERENCES 114

iii

APPENDICES A ESTIMATION OF CONCENTRATIONS IN VARIOUS SPECIES IN THE CATALYST PHASE A-1

B MODEL EQUATIONS (EQUALITY AND INEQUALITY CONSTRAINTS) A-13

C INPUT TO FLOWSHEET SIMULATION A-61 C.1 List of process u its A-63 C.2 List of process streams A-65 C.3 Measured values (i itial poi ts) A-69 C.4 U measured variables (i itial poi ts) A-72 C.5 Parameters A-105 C.6 Coefficie t values to calculate the e thalpy of gas phase A-107 C.7 Coefficie t values to calculate the e thalpy of liquid phase A-108 C.8 Co sta ts A-109

D OUTPUT FROM ON-LINE OPTIMIZATION A-111 D.1 Measured Variables A-111 D.2 U measured Variables A-129 D.3 Pla t Parameters A-331

E LIST OF STREAMS FROM THE ALKYLATION PROCESS MODEL USED FOR PINCH ANALYSIS A-343

F OUTPUT OF THE THEN PROGRAM A-345

G SEIP VALUES OF THE COMPONENTS IN THE ALKYLATION PROCESS A-351

H PROGRAM OUTPUTS A-352 H.1 Data Validatio Program (Do_data.lst) A-352 H.2 Parameter Estimatio Program (Do_para.lst) A-614 H.3 Eco omic Optimizatio Program (Do_eco .lst) A-881

I CALCULATION OF THE STEADY STATE OPERATION POINTS A-1141

J USER’S MANUAL A-1143

iv

ABSTRACT

Advanced Process Analysis System was successfully applied to the 15,000 BPD alkylation plant at the Motiva Enterprises Refinery in Convent, Louisiana. Using the flowsheeting, on-line optimization, pinch analysis and pollution assessment capabilities of the Advanced Process Analysis System an average increase in the profit of 127% can be achieved. Energy savings attained through reduced steam usage in the distillation columns total an average of 9.4x109 BTU/yr. A maximum reduction of 8.7% (67x109

BTU/yr) in heating and 6.0% (106x109 BTU/yr) in cooling requirements is shown to be obtainable through pinch analysis. Also one of the main sources of the pollution from the alkylation process, sulfuric acid consumption, can be reduced by 2.2 %.

Besides economic savings and waste reductions for the alkylation process, which is one of the most important refinery processes for producing conventional gasoline, the capabilities of an integrated system to facilitate the modeling and optimization of the process has been demonstrated. Application to other processes could generate comparable benefits.

The alkylation plant model had 1,579 equality constraints for material and energy balances, rate equations and equilibrium relations. There were 50 inequality constraints mainly accounting for the thermodynamic feasibility of the system. The equality and inequality constraints described 76 process units and 110 streams in the plant. Verification that the simulation described the performance of the plant was shown using data validation with 125 plant measurements from the distributed control system.

There were 1,509 unmeasured variables and 64 parameters in the model.

v List of Tables

s Table 2.1. Ψ k,l Values used in Alkylation Process Model 20

Table 2.2 Reaction Mechanism and Material Balances for Sulfuric Acid Alkylation of Isobutane with Propylene 32

Table 2.3 Reaction Mechanism and Material Balances for Sulfuric Acid Alkylation of Isobutane with Butylene 33

Table 4.1 Summary of the contractor model (5C-623) 71

Table 4.2 Summary of the acid settler model (5C-631) 72

Table 4.3 Summary of the depropanizer model (5C-603) 75

Table 4.4 Summary of the deisobutanizer model (5C-606A) 77

Table 4.5 Summary of the suction trap/flash drum model (5C-614) 81

Table 4.6 Summary of the compressor model (5K-601) 85

Table 4.7 Summary of the exchanger model (5E-628) 86

Table 4.8 Plant vs. Model Data 87

Table 5.1 Summary of Alkylation Model 94

Table 5.2 Measured Variables for operation point #1 95

Table 5.3 Plant Parameters for operation point #1 99

Table 5.4 Alkylation Plant Raw Material/Utility Costs and Product Prices 102

Table 5.5 Calculated Profit after Data Validation (D.V.), Parameter Estimation (P.E.) and Economic Optimization (E.O.) Steps for six Different Operation Points (Steady States) 103

Table 5.6 Input and Output Streams in Alkylation Process 109

Table 5.7. Pollution Assessment Values for Alkylation Process before (BEO) and after (AEO) the economic optimization. 110

vi List of Figures

Figure 1.1 Framework of Advanced Process Analysis System 2

Figure 1.2 Simplified Structure of On-Line Optimization 5

Figure 1.3 Reactor Design Program Outline 6

Figure 1.4 Composite Curves for Hot Streams and Cold Streams 8

Figure 1.5 Grid Diagram 8

Figure 2.1 Relationship between key elements of on-line optimization 15

Figure 2.2 Sulfuric Acid Alkylation Process (Vichailak 1995) 24

Figure 2.3 STRATCO Effluent Refrigeration Reactor (Yongkeat, 1996) 25

Figure 3.1 ‘Onion Skin’ Diagram for Organization of a Chemical Process and Hierarchy of Analysis 39

Figure 3.2 Example of Flowsim Screen for a Simple Refinery 42

Figure 4.1 Reactor and Refrigeration Sections of Alkylation Process 56

Figure 4.2 Depropanizer and Alkylate Deisobutanizer Sections of Alkylation Process 57

Figure 4.3 Saturate Deisobutanizer Section of Alkylation Process 58

Figure 4.4 STRATCO Effluent Refrigeration Reactor 59

Figure 4.5 Process flow diagram, as developed with the Flowsheet Simulation tool of Advanced Process Analysis System 65

Figure 4.6 Contactor 5C-623 66

Figure 4.7 Suction Trap Flash Drum (5C-614) 79

Figure 4.8 Economizer (5C-616) 82

Figure 4.9 Compressor (5K-601) 84

Figure 5.1 Grand Composite Curve for Alkylation Process 105

vii Figure 5.2 Network Grid Diagram for Alkylation Process 107

Figure 5.3 Integrating columns (5C-601 and 5C-603) with the process: Pressure shift for column 5C-601 only (left), for both columns (right). 108

viii

CHAPTE 1 INT ODUCTION This epo t documents the esults of applying the Advanced P ocess Analysis

System fo ene gy conse vation and pollution eduction in a comme cial, sulfu ic acid catalyzed, alkylation plant at the Motiva Ente p ises Refine y in Convent, Louisiana.

The Advanced P ocess Analysis System was developed fo use by p ocess and plant enginee s to pe fo m comp ehensive evaluations of p ojects in depth significantly beyond thei cu ent capabilities. The st ategy has the advanced p ocess analysis methodology identify sou ces of excess ene gy use and of pollutant gene ation. This p og am has built on esults f om esea ch on sou ce eduction th ough technology modification in eactions and sepa ations, ene gy conse vation (pinch analysis) and on- line optimization (p ocess cont ol) by P ofesso s Hoppe and Yaws at Lama and

P ofesso Pike at Louisiana State Unive sity. The System uses the Lama chemical

eacto analysis p og am, the LSU on-line optimization and pinch analysis p og ams, and the EPA pollution index methodology. Visual Basic was used to integ ate the p og ams and develop an inte active Windows inte face whe e info mation is sha ed th ough the Access database. This chapte gives an ove view of Advanced P ocess

Analysis System and an int oduction to the alkylation p ocess. These a e desc ibed in g eate detail subsequent chapte s.

1.1 Overview of the Advanced Process Analysis System

The advanced p ocess analysis methodology identifies sou ces of excess ene gy use and of pollutant gene ation was based on the f amewo k shown in Figu e 1.1. The main components of this system a e flowsheet simulation, on-line optimization, eacto analysis, pinch analysis, and pollution assessment. The flowsheet simulation p og am is used

1

Advanced Process Analysis System

On-Line Optimization Process Control Process Modification

Flowsheet eactor Pinch Pollution Simulation Analysis Analysis Index

Units, streams, physical property Flowsheet DataBase of APA : imulation data imulation PFD: units & streams Unit : local variables Units, streams, Process parameters physical property balance equations plant data On-Line pecification : stream connection PFD, units, streams, treams: global variables Optimal setpoints, Optimization physical properties reconciled data, Plant data parameters Property: enthalpy function density, viscosity Temp., flow rates enthalpy function Reactor F : simulation data Reactor comparison Analysis OLO: optimal setpoints reconciled data Temp., flow rates estimated parameters enthalpy function RA: reactor comparison Pinch Key word index: best reactor for the Best heat exchanger Analysis Unit ID, tream ID, process network Component ID, PA: best heat exchanger Property ID network Flow rates, composition PI: pollution information Pollution Pollution information Index

Figu e 1.1: F amewo k of Advanced P ocess Analysis System

2 fo p ocess mate ial and ene gy balances. Online optimization gives an accu ate desc iption of the chemical o efine y p ocess being evaluated. This p ocess simulation is used fo offline studies using eacto analysis, pinch analysis and pollution assessment, to achieve p ocess imp ovements that educe pollution and ene gy consumption.

The Advanced P ocess Analysis System has been applied to two contact p ocesses at the IMC Ag ico Company’s ag icultu al chemical complex. The esults of the application of the System showed a potential annual inc ease in p ofit of 3% (o

$350,000) and a 10% eduction in sulfu dioxide emissions ove cu ent ope ating conditions using the on-line optimization component of the System. The chemical

eacto analysis component showed that the eacto conve sion could be inc eased by

19% and that the eacto volume dec eased by 87% by using a eacto p essu e of 10.3 atm athe than the cu ent 1.3 atm. The pinch analysis component showed that the minimum amount of cooling wate was being used, and the heat exchange netwo k could be econfigu ed to educe the numbe of heat exchange s being used and educe the total heat exchange a ea by 25%. The pollution assessment component of the

System identified the sulfu fu nace and conve te s as the pa ts of the p ocess to be modified to minimize emissions. Details of these esults we e given in the thesis of

Keda Telang, 1998.

1.2 Flowsheet Simulation

The flowsheet simulation, Flowsim, is used to develop the p ocess model, and it has a g aphical use inte face with inte active capabilities. P ocess units a e ep esented as ectangula shapes whe eas the p ocess st eams a e ep esented as lines with a ows between these units. Each p ocess unit and st eam included in the flowsheet must have a

3 name and a desc iption. P ocess info mation is divided into the following six catego ies: equality const aints, inequality const aints, unmeasu ed va iables, measu ed va iables, pa amete s and constants. All of this p ocess info mation is ente ed with the help of the inte active, use -customized g aphic sc eens of Flowsim, and the info mation is sto ed in an Access database fo use by the othe p og ams.

The info mation in the fi st five catego ies is fu the classified by associating it with eithe a unit o a st eam in the flowsheet. Fo example, fo a unit that is a heat exchange , the elevant info mation includes the mass balance and heat t ansfe equations, limitations on the flow ates and tempe atu es if any, the heat t ansfe coefficient pa amete and all the inte mediate va iables defined fo that exchange .

Fo a st eam, the info mation includes its tempe atu e, p essu e, total flow ate, mola flow ates of individual components etc. Also, info mation not linked to any one unit o st eam is called the ‘Global Data’. Fo example, the ove all daily p ofit of the p ocess is aglobal unmeasu ed va iable.

The fo mulation of p ocess model fo the alkylation p ocess is desc ibed in detail in the use s’ manual in Appendix J. The on-line optimization p og am uses the p ocess model as const aint equations to maintain the p ocess ope ating at optimal set points in the dist ibuted cont ol system

1.3 On-line Optimization

Online optimization is the use of an automated system which adjusts the ope ation of a plant based on p oduct scheduling and p oduction cont ol to maximize p ofit and minimize emissions by p oviding setpoints to the dist ibuted cont ol system.

This is illust ated in Figu e 1.2. Plant data is sampled f om the dist ibuted cont ol system, and g oss e o s a e emoved f om it. Then, the data is econciled to be

4 consistent with the mate ial and ene gy balances of the p ocess. An economic model is used to compute the p ofit fo the plant and the plant model is used to dete mine the ope ating conditions, e.g. tempe atu es, p essu es, flow ates of the va ious st eams.

These a e va iables in the mate ial and ene gy balance of the plant model. The plant and economic model a e togethe used with an optimization algo ithm to dete mine the best ope ating conditions (e.g. tempe atu es, p essu es etc.) which maximizes the p ofit.

These optimal ope ating conditions a e then sent to the dist ibuted cont ol system to p ovide setpoints fo the cont olle s.

Figu e 1.2: Simplified St uctu e of Online Optimization

setpoints plant for measurements controllers

Distributed Control ystem sampled plant data

Gross Error optimal setpoint Detection operating targets and conditions Data Reconcilation

reconciled data

plant model Optimization Algorithm parameters Plant Model Economic Model Parameter Plant Model Estimation

economic model parameters

5

1.4 Chemical eactor Analysis

The Chemical Reacto Analysis p og am is a comp ehensive, inte active compute simulation fo th ee-phase catalytic gas-liquid eacto s and thei subsets, and an outline is shown in Figu e 1.3. The p og am has been developed by P ofesso

Hoppe and his esea ch g oup at Lama Unive sity (Saleh et al., 1995). It has a wide

ange of applications such as oxidation, hyd ogenation, hyd odesulfu ization, hyd oc acking and Fische -T opsch synthesis. This p og am inte actively guides the enginee to select the best eacto design fo the eacting system based on the cha acte istics of ten diffe ent types of indust ial catalytic gas-liquid eacto s which includes catalyst pa ticle diamete and loading, diffusivities, flow egimes, gas-liquid and liquid-solid mass t ansfe ates, gas and liquid dispe sions, heat t ansfe , holdup among othe s. The p og am solves the conse vation equations and has checks fo the validity of the design, e.g., not allowing a complete catalyst-wetting facto if the liquid flow ate is not sufficient. A mo e detailed desc iption is in the use 's manual.

Reaction

Homogeneous Hete ogeneous

Gas Phase Liquid Phase Catalytic Gas-Liquid

Gas Liquid Gas-Liquid CSTR Bubble Reacto PFR Packed Bed CSTR Batch Reacto Fixed Bed Reacto T ickle Bed Fluidised Bed Reacto Fixed Bubble Bed CSTR Slu y Bubble Slu y 3-Phase Fluidised Bed

Figu e 1.3: The Reacto Design P og am Outline

6

1.5 Pinch Analysis

Pinch technology was developed in the late 1970's as a method fo the design of heat exchange netwo ks, and it has since been extended to site ene gy integ ation including distillation and utility systems, mass exchange s, and a numbe of othe applications (Linnhoff, 1993; Gupta and Manousiouthakis, 1993). Pinch analysis dete mines the best design fo sepa ations, ecycle and heat exchange netwo ks. It employs th ee concepts: the composite cu ves, the g id diag am of p ocess st eams and the pinch point; and these a e applied to minimize ene gy use in the p ocess.

Illust ations of composite cu ves and the g id diag am a e shown in Figu e 1.4 and

Figu e 1.5 espectively. The composite cu ves a e plots of tempe atu e as a function of enthalpy f om the mate ial and ene gy balances fo the st eams that need to be heated, called cold st eams, and those that need to be cooled, called hot st eams. F om the composite cu ves of the hot and cold st eams, the potential fo ene gy exchange between the hot and cold st eams can be dete mined, as well as the p ocess

equi ements fo exte nal heating and cooling f om utilities such as steam and cooling wate . At one o mo e points the cu ves fo the hot and cold st eams may come ve y close, the p ocess pinch; and this means the e is no su plus heat fo use at lowe tempe atu es. The g id diag am has ve tical lines to ep esent the hot and cold st eams with lengths co esponding to the tempe atu e ange with the hot st eams going f om top left and the cold st eams f om bottom ight. With this a angement the heat

ecove y netwo k fo the p ocess design can be dete mined. A g and composite, tempe atu e-enthalpy cu ve can be assembled f om the composite cu ves and the g id diag am to help select utilities and app op iately place boile s, tu bines, distillation columns, evapo ato s and fu naces. Also, the heat t ansfe su face a ea can be

7 dete mined with the co esponding capital cost fo both ene gy and cost minimization.

This methodology is inco po ated in compute p og am THEN which is inco po ated in the Advanced P ocess Analysis System.

160 160

120 120

T (°C) 80 T (°C) 80 C1+C2 H1+H2 40 40

0 0 0 100 200 300 400 500 0 100 200 300 400 500 Q (W) Q (W)

Figu e 1.4: Composite Cu ves fo Hot St eams (on the left side) and Cold St eams (on the ight side) fo the Simple P ocess

H1 1 4 1

H2 2 2

3

1 C1

2 C2

Heate Coole Heat Exchange Loop

Figu e 1.5: G id Diag am in selecting utilities and app op iate placement of boile s, tu bines, distillation columns, evapo ato s and fu naces.

8

1.6 Pollution Assessment

The pollution assessment module of the Advanced P ocess Analysis System is called ‘The Pollution Index P og am’. It is based on the Waste Reduction Algo ithm

(WAR) (Hilaly, 1994) and the Envi onmental Impact Theo y (Cabezas et. al., 1997).

The WAR algo ithm is based on the gene ic pollution balance of a p ocess flow diag am.

Pollution Accumulation = Pollution Inputs +

Pollution Gene ation -Pollution Output (1)

It defines a quantity called as the 'Pollution Index' to measu e the waste gene ation in the p ocess. This pollution index is defined as:

I = wastes/p oducts = -(ΓOut + ΓFugitive) / ΓPn (2)

This index is used to identify st eams and pa ts of p ocesses to be modified.

Also, it allows compa ison of pollution p oduction of diffe ent p ocesses. The WAR algo ithm can be used to minimize waste in the design of new p ocesses as well as modification of existing p ocesses.

The Envi onmental Impact Theo y (Cabezas et. al., 1997) is a gene alization of the WAR algo ithm. It desc ibes the methodology fo evaluating potential envi onmental impacts, and it can be used in the design and modification of chemical p ocesses. The envi onmental impacts of a chemical p ocess a e gene ally caused by the ene gy and mate ial that the p ocess takes f om and emits to the envi onment. The potential envi onmental impact is a conceptual quantity that can not be measu ed. But it can be calculated f om elated measu able quantities.

9

1.7 Alkylation

Alkylation is an impo tant pet oleum efining p ocess that is used to conve t light isopa affins and light olefins into high octane numbe isopa affins. Isopa affins containing a te tia y ca bon atom unde go catalytic alkylation with C3-C5 olefins to p oduce highly b anched pa affins in the C7-C9 ange. This involves a composite of consecutive and simultaneous eactions including polyme ization, disp opo tionation, c acking and self-alkylation eactions (Co ma and Ma tinez, 1993). Comme cially, isobutane is used fo the p ocess because isopentane and highe isopa affins have octane numbe s that a e quite desi able.

Catalytic alkylation occu s in the p esence of sulfu ic (H2SO4) o hyd ofluo ic acid (HF) catalysts, at mild tempe atu es and at sufficient p essu e to maintain the hyd oca bons in the liquid state. With sulfu ic acid it is necessa y to ca y out the

eactions at 10 to 20°C (50 to 70°F) o lowe , to minimize oxidation- eduction

eactions, which esult in fo mation of ta s and p oduction of sulfu dioxide. When hyd ofluo ic acid is the catalyst, eaction tempe atu e is usually limited to 35°C (100°F) o lowe . The catalyst exists as a sepa ate phase, and the eactants and p oducts must be t ansfe ed to and f om the catalyst (G use and Stevens 1960, Rosenwald 1978).

Comme cial alkylation plants use eithe sulfu ic acid (H2SO4) o hyd ofluo ic acid (HF) as catalysts. About 20 yea s ago almost th ee times as much alkylate was p oduced using H2SO4 as the catalyst as compa ed to p ocesses using HF. Since then the

elative impo tance of p ocesses using HF has inc eased substantially and cu ently these p ocesses p oduce in the U.S. about 47% of the alkylate. Howeve , in the last five yea s, mo e H2SO4 than HF type units have been built due to envi onmental and safety conce ns. Recent info mation cla ifying the dange s of HF is causing efine ies that use

10

HF to econside the catalyst, o imp ove the safety of equipment and p ocedu es

(Alb ight 1990a, Cupit et al. 1961).

1.8 Summary

An ove view of the Advanced P ocess Analysis System was given and successful applications to othe p ocesses we e desc ibed b iefly. Also, the cu ent status of comme cial alkylation p ocesses was given.

In the next chapte a lite atu e su vey is given on cu ent status of the techniques inco po ated in the Advanced P ocess Analysis System and alkylation p ocess chemist y and ope ations. The thi d chapte of this epo t desc ibes the application of the

Advanced P ocess Analysis System to the alkylation p ocess. The fou th chapte gives the desc iption of the development of the p ocess model of the alkylation p ocess using

Advanced P ocess Analysis System. The fifth chapte p esents the esults of the analysis fo the p ocess.

11

CHAPTE 2

LITE ATU E EVIEW In
t is
c apter
t e
current
status
of
t e
met odology
and
literature
is
reviewed
for
t e
met ods
used
in
t e
Advanced
Process
Analysis
System.
Also
t e
current
understanding
of
t e
alkylation
process
and
its
tec nology
is
given.

2.1 Advanced Process Analysis System

Advanced
Process
Analysis
System,
based
on
t e
framework
given
in
Figure

1.1,
includes
c emical
reactor
analysis,
process
flows eeting,
pinc
analysis
and
on!line
optimization.
All
of
t ese
programs
use
t e
same
information
of
c emical
processes

(material
and
energy
balances,
rate
equations
and
equilibrium
relations).

Consequently,
an
advanced
and
integrated
approac
for
process
analysis
is
available
now.

T e
need
of
an
integrated
approac
to
process
analysis
as
been
given
by
Van

Reeuwijk
et
al.
(1993)
w o
proposed
aving
a
team
of
computer
aided
process
engineering
expert
and
a
process
engineer
wit
tec nology
knowledge
to
develop
energy
efficient
c emical
processes.
A
process
engineer
software
environment
is
described
by
Ballinger
et
al.
(1994)
called
‘epee’
w ose
goal
is
ave
to
a
user
interface
to
create
and
manipulate
objects
suc
as
processes,
streams
and
components
wit
s aring
of
data
among
process
engineering
applications
in
an
open
distributed
environment.
T e
Clean
Process
Advisory
System
(CPAS)
as
been
described
by
Baker
et
al.(1995)
as
a
computer
based
pollution
prevention
process
and
product
design
system
t at
contains
ten
PC
software
tools
being
developed
by
an
industry!government! university
team.

T is
includes
tec nology
selection
and
sizing,
potential
and
designs,
p ysical
property
data,
materials
locators
and
regulatory
guidance
information.
An
article
by
S aney
(1995)
describes
t e
various
modeling
software
and
databases

12

available
for
process
analysis
and
design.
A
review
of
computer
aided
process
engineering
by
Winter
(1992)
predicts
linking
various
applications
will
result
in
better
quality
of
process
design,
better
plant
operations
and
increased
productivity.
It
also
describes
t e
PRODABAS
concept,
w ic
focuses
on
capturing
information
from
multiple
sources
into
a
common
multi!user
framework
for
analysis,
process
definition
and
process
engineering
documentation
rat er
t an
t e
original
concept
of
a
common
user
interface
and
datastore
linked
wit
a
range
of
applications
computing
tools.

2.1.1 Industrial Applications of On-Line Optimization

Boston,
et
al.,
(1993)
gave
a
wide
review
of
computer
simulation
and
optimization
as
well
as
advanced
control
in
c emical
process
industries.
He
described
t e
new
computing
power
for
process
optimization
and
control
t at
leads
to
ig er
product
qualities
and
better
processes,
w ic
are
cleaner,
safer,
more
efficient,
and
less
costly.

Lauks,
et
al.,
(1992)
reviewed
t e
industrial
applications
of
on!line
optimization
reported
in
t e
literature
from
1983
to
1991
and
cited
nine
applications
–
five
et ylene
plants,
a
refinery,
a
gas
plant,
a
crude
unit
and
a
power
station.
T e
results
s owed
a
profitability
increase
of
3%
or
$4M/year.
Also,
intangible
profits
from
a
better
understanding
of
plant
be avior
were
significant.

Z ang
(1993)
conducted
a
study
of
on!line
optimization
for
Monsanto!
designed
sulfuric
acid
plant
of
IMC
Agrico
at
Convent,
Louisiana.
Economic
optimization
can
ac ieve
a
17%
increase
in
plant
profit
and
25%
reduction
in
sulfur
dioxide
emission.

T e
plant
was
studied
by
C en
(1998),
in
developing
t e
optimal
way
to
conduct
on!

13

line
optimization.
Also,
C en
(1998)
reported
a
number
of
ot er
successful
applications
of
on!line
optimization
in
improving
c emical
processes.

2.1.2 Key Elements of On-Line Optimization

T e
objective
of
on!line
optimization
is
to
determine
optimal
process
setpoints
based
on
plant’s
current
operating
and
economic
conditions.
As
s own
in
Figure
1.2,
t e
key
elements
of
on!line
optimizations
are
(C en,
1998):

! Gross
Error
Detection

! Data
Reconciliation

! Parameter
Estimation

! Economic
Model
(Profit
Function)

! Plant
Model
(Process
Simulation)

! Optimization
Algorit m

T e
procedure
for
implementing
on!line
optimization
involves
steady!state
detection,
data
validation,
parameter
estimation,
and
economic
optimization.

T e
relations ip
between
t ese
key
elements
is
outlined
in
Figure
2.1.
Bot
plant
model
and
optimization
algorit ms
are
required
in
t e
t ree
steps
of
on!line
optimization
–
data
validation,
parameter
estimation,
and
economic
optimization.
Plant
model
serves
as
constraint
equations
in
t e
t ree
nonlinear
optimization
problems,
w ic
are
solved
by
t e
optimization
algorit m.
For
data
validation,
errors
in
plant
measurements
are
rectified
by
optimizing
a
likeli ood
function
subject
to
plant
model,
and
a
test
statistic
is
used
to
detect
gross
errors
in
t e
measurements.
For
parameter
estimation,
parameters
in
plant
model
are
estimated
by
optimizing
an
objective
function,
suc
as
minimizing
t e
sum
of
squares
of
measurement
errors,
subject
to

14

constraints
in
t e
plant
model.
For
economic
optimization,
t e
plant
model
is
used
wit
economic
model
to
maximize
plant
profit
and
provide
optimal
setpoints
for
t e
distributed
control
system
to
operate.

Figure
2.1
Relations ip
between
key
elements
of
on!line
optimization

2.1.3 Energy Conservation

Heat
Exc anger
Network
Synt esis
(HENS)
for
maximum
eat
recovery
is
t e
key
to
energy
conservation
in
a
c emical
plant.
T e
problem
of
design
and
optimization
of
eat
exc anger
networks
as
received
considerable
attention
over
t e
last
two
decades
(A mad
et
al.,
1990,
Duran
et
al.,
1986,
Lin off
et
al.
1978,
1979,
1982).

T e
problem
of
HENS
can
be
defined
as
t e
determination
of
a
cost!effective
network
to
exc ange
eat
among
a
set
of
process
streams
w ere
any
eating
and
cooling
t at
is
not
satisfied
by
exc ange
among
t ese
streams
must
be
provided
by
external
utilities
(S enoy,
1995).
Attempts
at
solving
t is
problem
ave
been
based
on
t e
following
approac es.

Heuristic Approaches:
T e
HENS
problem
was
first
formulated
by
Masso
and

Rudd
(1969).
At
t at
time,
t e
design
met ods
were
generally
based
on
euristic

15

approac es.
One
of
t e
commonly
used
rules
was
to
matc
t e
ottest
stream
wit
t e
coldest
stream.
Several
ot er
met ods
were
based
on
tree
searc
tec niques
(Lee
et
al.,

1970).
T is
generally
led
to
feasible
but
non!optimal
solutions.

Pinch Analysis: Ho mann
(1971)
made
significant
contributions
to
t e
development
of
t e
t ermodynamic
approac .
In
t e
late
1970s,
Linn off
and

Hindmars
(Linn off
et
al.,
1979)
first
introduced
Pinc
Analysis;
a
met od
based
on
t ermodynamic
principles.
T ey
also
introduced
a
number
of
important concepts, w ic
formed
t e
basis
for
furt er
researc .
T ese
concepts
were
reviewed
by
Gundersen
et
al.

(1987),
and
summarized
by
Telang
(1998).

Mathematical Programming: Developments
in
computer
ardware
and
software
enabled
t e
development
of
met ods
based
on
mat ematical
programming.

Paoulias
and
Grossman
(1983)
formulated
Maximum
Energy
Recovery
(MER)
problem
as
a
linear
programming
(LP)
model
based
on
t e
transs ipment
model,
w ic
is
widely
used
in
operations
researc .
T is
model
was
expanded
to
make
restricted
ot
and
cold
stream
matc es
by
using
mixed
integer
linear
programming
(MILP)
formulation.

HEXTRAN,
SUPERTARGET
and
ASPEN
PINCH
are
some
of
t e
commonly
used
commercial
eat
exc anger
design
programs.

2.1.4 Pollution Prevention

Cost
minimization
as
traditionally
been
t e
objective
of
c emical
process
design.
However,
growing
environmental
awareness
now
demands
process
tec nologies
t at
minimize
or
prevent
production
of
wastes.
T e
most
important
issue
in
development
of
suc
tec nologies
is
a
met od
to
provide
a
quantitative
measure
of
waste
production
in
a
process.

16

Waste eduction Algorithm:
Many
different
approac es
(Telang,
1998)
ave
been
suggested
to
deal
wit
t is
problem.
One
of
t ese
is
t e
Waste
Reduction

Algorit m
(WAR)
(Hilaly,
1994).
T e
WAR
algorit m
is
based
on
t e
generic
pollution
balance
of
a
process
flow
diagram.

Pollution
Accumulation
=
Pollution
Inputs
+







Pollution
Generation
!
Pollution
Output


(2.1)

It
defines
a
quantity
called
as
t e
'Pollution
Index'
to
measure
t e
waste
generation
in
t e
process.
T is
pollution
index
is
defined
as:

I
=
wastes/products
=
!
(ΓOut
+
ΓFugitive)
/
ΓPn


(2.2)

T is
index
is
used
to
identify
streams
and
parts
of
processes
to
be
modified.

Also,
it
allows
comparison
of
pollution
production
of
different
processes.
T e
WAR
algorit m
can
be
used
to
minimize
waste
in
t e
design
of
new
processes
as
well
as
modification
of
existing
processes.

Environmental Impact Theory: T is
t eory
(Cabezas
et.
al.,
1997)
is
a
generalization
of
t e
WAR
algorit m.
It
describes
t e
met odology
for
evaluating
potential
environmental
impacts,
and
it
can
be
used
in
t e
design
and
modification
of
c emical
processes.
T e
environmental
impacts
of
a
c emical
process
are
generally
caused
by
t e
energy
and
material
t at
t e
process
takes
from
and
emits
to
t e
environment.

T e
potential
environmental
impact
is
a
conceptual
quantity
t at
can
not
be
measured.
But
it
can
be
calculated
from
related
measurable
quantities.

T e
generic
pollution
balance
equation
of
t e
WAR
algorit m
is
now
applied
to
t e
conservation
of
t e
Potential
Environmental
Impact
in
a
process.
T e
flow
of
impact

I& ,
in
and
out
of
t e
process
is
related
to
mass
and
energy
flows
but
is
not
equivalent
to
t em.
T e
conservation
equation
can
be
written
as

17

dIsys = I&& n − Iout + I&gen dt (2.3)

w ere I sys
is
t e
potential
environmental
impact
content
inside
t e
process,
I& n
is
t e

input
rate
of
impact,
I&out
is
t e
output
rate
of
impact
and
I&gen
is
t e
rate
of
impact
generation
inside
t e
process
by
c emical
reactions
or
ot er
means.
At
steady
state,
equation
2.3
reduces
to:

0 = I&& n − Iout + I&gen



(2.4)

Application
of
t is
equation
to
c emical
processes
requires
an
expression
t at
relates
t e
conceptual
impact
quantity
I&
to
measurable
quantities.
T e
input
rate
of
impact
can
be
written
as

n
I = Ij = Mj xkjΨ & n ∑ &&∑ ∑ k (2.5)
j j k w ere
t e
subscript
‘in’
stands
for
input
streams.
T e
sum
over
j
is
taken
over
all
t e
input
streams.
For
eac
input
stream
j,
a
sum
is
taken
over
all
t e
c emical
species
present
in
t at
stream.
Mj
is
t e
mass
flow
rate
of
t e
stream
j
and
t e
xkj
is
t e
mass
fraction
of
c emical
k
in
t at
stream.
Ψk
is
t e
c aracteristic
potential
impact
of
c emical
k.

T e
output
streams
are
furt er
divided
into
two
different
types:
Product
and

Non!product.
All
non!product
streams
are
considered
as
pollutants
wit
positive
potential
impact
and
all
product
streams
are
considered
to
ave
zero
potential
impact.

18

T e
output
rate
of
impact
can
be
written
as:

= = out Ψ
I&out ∑ I&&j ∑ Mj ∑ xkj k j j k (2.6)
w ere
t e
subscript
‘out’
stands
for
non!product
streams.
T e
sum
over
j
is
taken
over
all
t e
non!product
streams.
For
eac
stream
j,
a
sum
is
taken
over
all
t e
c emical
species.

Knowing
t e
input
and
output
rate
of
impact
from
t e
equations
2.5
and
2.6,
t e
generation
rate
can
be
calculated
using
equation
2.4.
Equations
2.5
and
2.6
need
values
of
potential
environmental
impacts
of
c emical
species.
T e
potential
environmental

impact
of
a
c emical
species
( Ψk )
is
calculated
using
t e
following
expression

∑ s
(2.7)
Ψ k = Ψ l l k, l w ere,
t e
sum
is
taken
over
t e
categories
of
environmental
impact.
αl
is
t e
relative

s weig ting
factor
for
impact
of
type
l
independent
of
c emical
k.
Ψ k,l
l
(units
of

Potential
Environmental
Impact/mass
of
c emical
k)
is
t e
potential
environmental

s impact
of
c emical
k
for
impact
of
type
l.
Values
of
Ψ k,l
for
a
number
of
c emical
species
can
be
obtained
from
t e
report
on
environmental
life
cycle
assessment
of

s products
(Heijungs,
1992).
Some
non!zero
values
of
Ψ k,l
for
t e
components
used
in
t e
modeling
of
t e
alkylation
process
are
given
in
Table
2.1.

T ere
are
nine
different
categories
of
impact.
T ese
can
be
subdivided
into
four
p ysical
potential
impacts
(acidification,
green ouse
en ancement,
ozone
depletion
and
p otoc emical
oxidant
formation),
t ree
uman
toxicity
effects
(air,
water
and
soil)
and
two
ecotoxicity
effects
(aquatic
and
terrestrial).
T e
relative
weig ting
factor
αl
allows
t e
above
expression
for
t e
impact
to
be
customized
to
specific
or
local
conditions.

T e
suggested
procedure
is
to
initially
set
values
of
all
relative
weig ting
factors
αl
to

19

one,
and
t en
allow
t e
user
to
vary
t em
according
to
local
needs.
More
information
on
impact
types
and
c oice
of
weig ting
factors
can
be
obtained
from
t e
report
on
environmental
life
cycle
assessment
of
products
(Heijungs,
1992).

s Table
2.1.
Ψ k,l
Values
used
in
Alkylation
Process
Model

Component

Ecotoxicity
Ecotoxicity
Human
Human
Human
P otoc emical
(aquatic)
(terrestrial)
Toxicity
Toxicity
Toxicity
Oxidant
(air)
(water)
(soil)
Formation
C3 - 0.0305
0
9.06E!7
0
0
1.1764
C4 = 0.0412
0.3012
0
0.3012
0.3012
1.6460
iC4 0.1566
0.2908
8.58E!7
0.2908
0.2908
0.6473
nC4 0.1890
0.2908
8.58E!7
0.2908
0.2908
0.8425
iC5 0.0649
0.2342
0
0.2342
0.2342
0.6082
nC5 0.3422
0.2342
5.53E!7
0.2342
0.2342
0.8384
iC6 0.2827
0.1611
0
0.1611
0.1611
1.022
H2SO4 0.0170
0.1640
0.2950
0.1640
0.1640
0

To
quantitatively
describe
t e
pollution
impact
of
a
process,
t e
conservation
equation
is
used
to
define
two
categories
of
Impact
Indexes.
T e
first
category
is
based
on
generation
of
potential
impact
wit in
t e
process.
T ese
are
useful
in
addressing
t e
questions
related
to
t e
internal
environmental
efficiency
of
t e
process
plant,
i.e.,
t e
ability
of
t e
process
to
produce
desired
products
w ile
creating
a
minimum
of
environmental
impact.
T e
second
category
measures
t e
emission
of
potential
impact
by
t e
process.
T is
is
a
measure
of
t e
external
environmental
efficiency
of
t e
process
i.e.
t e
ability
to
produce
t e
desired
products
w ile
inflicting
on
t e
environment
a
minimum
of
impact.

Wit in
eac
of
t ese
categories,
t ree
types
of
indexes
are
defined
w ic
can
be
used
for
comparison
of
different
processes.
In
t e
first
category
(generation),
t e
t ree
indexes
are
as
follows.

20

NP 1) I&gen
T is
measures
t e
t e
total
rate
at
w ic
t e
process
generates
potential

environmental
impact
due
to
nonproducts.
T is
can
be
calculated

by

subtracting
t e
input
rate
of
impact
( I& n )
from
t e
output
rate
of
impact

NP ( I&out ),
i.e.
I&gen
=
I&out ! I& n .

NP 2)

I$gen
T is
measures
t e
potential
impact
created
by
all
nonproducts
in

manufacturing
a
unit
mass
of
all
t e
products.
T is
can
be
obtained
from

NP
dividing

I&gen
by
t e
rate
at
w ic
t e
process
outputs
products,
i.e.

− NP I gen NP

I$gen
=

− . ∑ Pp p

NP 3) M$ gen
T is
is
a
measure
of
t e
mass
efficiency
of
t e
process,
i.e.,
t e
ratio
of

mass
converted
to
an
undesirable
form
to
mass
converted
to
a
desirable

NP form.
T is
can
be
calculated
from
I$gen
by
assigning
a
value
of
1
to
t e

potential
impacts
of
all
non!products,
i.e.

− (out) − ( n) NP − NP ∑ M j ∑ xkj ∑ M j ∑ xkj j k j k NP

M$ gen
=
− . ∑ Pp p

T e
indexes
in
t e
second
category
(emission)

are
as
follows.

NP 4) I&out
T is
measures
t e
t e
total
rate
at
w ic
t e
process
outputs
potential

environmental
impact
due
to
nonproducts.
T is
is
calculated
using
equation

− (out) NP NP Ψ 2.6,
i.e.
I&out
=
∑ M j ∑ x kj k .
j k

21

NP 5) I$out
T is
measures
t e
potential
impact
emitted
in
manufacturing
a
unit
mass
of

NP all
t e
products.
T is
is
obtained
from
dividing
I&out
by
t e
rate
at
w ic
t e

− NP NP I out process
outputs
products,
i.e.
I$out =
− .
∑ Pp p

NP 6) M$ out
T is
is
t e
amount
of
pollutant
mass
emitted
in
manufacturing
a
unit
mass

NP of
product.
T is
can
be
calculated
from
I$out
by
assigning
a
value
of
1
to
t e

potential
impacts
of
all
non!products,
i.e.

− (out) NP ∑ M j ∑ xkj NP j k
M$ out =
− .
∑ Pp p

Indices
1
and
4
can
be
used
for
comparison
of
different
designs
on
an
absolute
basis
w ereas
indices
2,
3,
5
and
6
can
be
used
to
compare
t em
independent
of
t e
plant
size.
Hig er
values
of
indices
mean
ig er
pollution
impact
and
suggest
t at
t e
plant
design
is
inefficient
from
environmental
safety
point
of
view.
Negative
values
mean
t at
t e
input
streams
are
actually
more
armful
to
t e
environment
t an
t e
non! products
if
t ey
are
not
processed.

2.2 Sulfuric Acid Alkylation Process

Alkylation
offers
several
key
advantages
to
refiners,
including
t e
ig est
average
quality
of
all
components
available
to
t e
gasoline
pool,
increased
amounts
of
gasoline
per
volume
of
crude
oil
and
ig
eats
of
combustion.
Alkylates
permit
use
of
internal
combustion
engines
wit
ig er
compression
ratios
and
ence
t e
potential
for

22

increased
miles
per
gallon.
Alkylates
burn
freely,
promote
long
engine
life,
and
ave
low
levels
of
undesired
emissions
(Albrig t
1990a,
Corma
and
Martinez
1993).

T e
catalytic
alkylation
of
paraffins
involves
t e
addition
of
an
isoparaffin
containing
tertiary
ydrogen
to
an
olefin.
T e
process
is
used
by
t e
petroleum
industry
to
prepare
ig ly
branc ed
paraffins
mainly
in
t e
C7
to
C9
range
for
use
as
ig !quality
fuels
for
spark
ignition
engines.
T e
overall
process
is
a
composite
of
complex
reactions,
and
consequently
rigorous
control
is
required
of
operating
conditions
and
of
catalyst
to
assure
predictable
results.

2.2.1 Alkylation in the Petroleum Industry

Isoparaffin!olefin
alkylation
entails
t e
manufacture
of
branc ed
paraffins
t at
distill
in
t e
gasoline
range
(up
to
ca.
200
oC).
Commercial
refinery
plants
operate
wit
t e
C3
and
C4
ydrocarbon
streams;
alkylation
involving
ig
molecular
weig t
olefin
or
isoparaffins
(over
C5)
are
not
attractive,
partly
because
of
numerous
side
reactions
suc
as
ydrogen
transfer
(Rosenwald
1978).

Sulfuric
acid
concentration
is
maintained
at
about
90%.
Operation
below
t is
acid
concentration
generally
causes
polymerization.
Product
quality
is
improved
w en
temperatures
are
reduced
to
t e
range
of
0!10
oC.
Cooling
requirements
are
obtained
by
flas ing
of
unreacted
isobutane.
Some
form
of
eat
removal
is
essential
because
t e

eat
of
reaction
is
approximately
14
x
105
J/kg
(600
Btu/lb.)
for
butenes.
In
order
to
prevent
polymerization
of
t e
olefin,
an
excess
of
isobutane
is
c arged
to
t e
reaction
zone.
Isobutane!to!olefin
molar
ratios
of
6:1
to
14:1
are
common.
More
effective
suppression
of
side
reactions
is
produced
by
t e
ig er
ratios
(Vic ailak
1995).

23

T e
alkylation
reaction
system
is
a
two!p ase
system
wit
a
low
solubility
of
isobutane
in
t e
catalyst
p ase.
In
order
to
ensure
intimate
contact
of
t e
reactant
and
t e
catalyst,
efficient
mixing
wit
fine
subdivision
must
be
provided.
Presence
of
unsaturated
organic
diluent
in
t e
acid
catalyst
favors
t e
alkylation
reaction.
T e
organic
diluent
as
been
considered
to
be
a
source
of
carbonium
ions
t at
promote
t e
alkylation
reaction
(Rosenwald
1978).

2.2.2 Commercial Sulfuric Acid Alkylation Process

More
t an
60%
of
t e
worldwide
production
of
alkylate
using
a
sulfuric
acid
catalyst
is
obtained
from
effluent
refrigeration
process
of
Stratco
Inc.
A
typical
process
flow
diagram
is
as
s own
in
Figure
2.2.

Figure
2.2.
Sulfuric
Acid
Alkylation
Process
(Vic ailak
1995)

24

T e
Stratco
reactor
or
contactor,
s own
in
Figure
2.3,
is
a
orizontal
pressure
vessel
containing
a
mixing
impeller,
an
inner
circulation
tube
and
a
tube
bundle
to
remove
t e
eat
generated
by
t e
alkylation
reaction.
T e
ydrocarbon
and
acid
feeds
are
injected
into
t e
suction
side
of
t e
impeller
inside
t e
circulation
tube.
T e
impeller
rapidly
disperses
t e
ydrocarbon
feed
wit
t e
acid
catalyst
to
form
an
emulsion.
T e
emulsion
is
circulated
by
t e
impeller
at
ig
rates
wit in
t e
contactor.

Figure
2.3.
STRATCO
Effluent
Refrigeration
Reactor
(Yongkeat,
1996)

Stratco
contactors
are
usually
sized
to
produce
2,000
bbl/day
of
alkylate

(Albrig t
1990a).
Improvements
introduced
in
recent
years
include:
longer
coils
to
increase
t e
overall
eat
transfer
coefficients;
improved
pump/agitator
system
and
injection
devices
for
introducing
t e
ydrocarbon
feed
and
t e
acid
into
t e
contactor,
w ic
in
turn
improves
alkylate
quality
and
lowers
refrigeration
costs.
T e
pump/agitator
as
been
positioned
below
t e
centerline
in
order
to
minimize
partial
settling
of
acid
at
t e
bottom
of
t e
contactor.

A
part
of
t e
emulsion
is
continuously
removed
and
sent
to
t e
acid
settler
or
decanter,
w ere
t e
acid
and
ydrocarbon
p ases
separate.
T en
in
a
flas
drum
t e

ydrocarbon
p ase
is
flas ed
to
separate
C4
and
lig ter
components
from
t e
eavier

25

ydrocarbons.
T e
lig ter
components
are
sent
to
t e
contactor
as
refrigerants
and
are
t en
recycled.
T e
eavier
components
are
sent
to
t e
deisobutanizer
column,
w ere
alkylate
is
separated
from
unreacted
butane
and
isobutane.
T e
C3
content
in
t e
system
is
decreased
by
sending
t e
lig ter
components
t roug
t e
depropanizer
column.
T e
process
is
described
in
greater
detail
in
C apter
4.

2.2.3 Theory of Alkylation eactions

Alkylation
of
isobutane
wit
C3!C5
olefins
involves
a
series
of
consecutive
and
simultaneous
reactions
(Corma
and
Martinez
1993).
Only
isoparaffins
containing
a
tertiary
carbon
atom
are
found
to
undergo
catalytic
alkylation
wit
olefins.
Reactions
and
products
are
readily
explained
by
t e
carbonium
ion
mec anism.

T e
principal
reactions
t at
occur
in
alkylation
are
t e
combinations
of
olefins
wit
isoparaffins
as
follows:

CH3






















CH3


























CH3








CH3

|




























|
































|















|
CH3
!
C
=
CH2

+

CH3
!
CH
!
CH3

→

CH3
!
C
!
CH2
!
CH
!
CH3









































































|

CH3

isobutylene

isobutane








2,2,4
!
trimet ylpentane
















































































(isooctane)














































CH3




























CH3













































|
































|
CH2
=
CH
!
CH3

+

CH3
!
CH
!
CH3

→

CH3
!
CH
!
CH2
!
CH2
!
CH3















































































|














































































CH3



propylene

















isobutane
























2,2
!
dimet ylpentane





















































































(iso eptane)

26

Steps
in
t e
alkylation
reaction
mec anism
involving
carbonium
ions
are
s own
below,
wit
typical
examples
to
illustrate
eac
reaction
step
(Cupit
1961).
(X
is
OSO3H
or
F):

1. T e
first
step
is
t e
addition
of
proton
to
olefin
molecule
to
form
a
tertiary
butyl

cation:

C





























X













|
































|





















+





























C
!
C
=
C

+

HX




C!
C
!
C




C
!
C
!
C

+

X!











































(2.8)









































|





















|








































C



















C
2. T en,
t e
tertiary
butyl
cation
is
added
to
t e
olefin
:

C
















|










+
C
=
C
!
C










→

C
!
C
!
C
!
C
!
C













(2.9)








|






























|










|







C




























C








C
















C







C








|



















|










+

C
–
C+

+
C
!
C
=
C
!
C




→

C
!
C
!
C
!
C
!
C



































(2.10)








|



















|




|







C







C


C













C










|










+



C
=
C
!
C
!
C




→

C
!
C
!
C
!
C
!
C
!
C












(2.11)










|









C

27

3. Carbonium
ions
may
isomerize
via
ydride
and
met yl
s ifts
to
form
more
stable

carbonium
ions:

C


C



C
!









































~CH3






|




|





|
→ C
!
C
!
C
!
C
!
C





















































(2.12)





















+














C



C































C


C





























C



C
! !













|





|




+
















~H








|




|















~CH3









|





|

C
!
C
!
C
!
C
!
C


→


C
!
C
!
C
!
C
!
C



→



C
!
C
!

C
!
C
!
C









(2.13)














|







































|




+






























+







|













C





































C








C























































C








C










C








C
! !









































~CH3







|










|







~H










|










|








→


C
!
C
!
C
!
C
!
C



→



C
!
C
!
C
!
C
!
C









(2.14)





















|




+











|










+






























C











C

4. T ese
carbonium
ions
suffer
rapid
ydride
transfer
from
isobutane,
leading
to
t e

different
paraffin
isomers
and
generating
tertiary
butyl
cation:

C



C


C

















C









C


C


C






C








|





|




|










|










|




|




|







|
C
!
C
!
C
!
C
!
C

+

C
!
C
!
C

→

C
!
C
!
C
!
C
!
C

+

C
!
C
!
C



















(2.15)


+












+

Unfortunately,
t ese
are
not
t e
only
reactions
occurring
during
alkylation.
T ere

are
a
number
of
secondary
reactions
t at
in
general
reduce
t e
quality
of
t e

alkylate.
T ese
reactions
include
polymerization,
disproportionation,
cracking
and

self!alkylation
reactions
(Corma
and
Martinez
1993).

5. Polymerization
results
from
t e
addition
of
a
second
olefin
to
t e
carbonium
ion

formed
in
t e
primary
reaction,
as
was
seen
in
step
2
of
t e
above
mec anism:

+ + + iC8H17

+
C4H8

→
iC12H25

+
C4H10

→
iC12H26

+
iC4H9











(2.16)

28

+ T e
iC12H25
can
continue
to
react
wit
an
olefin
to
form
a
larger
isoalkyl
cation:

+ + iC12H25

+
C4H8

→
iC16H33

+
iC4H10

→
iC16H34

+
iC4H9











(2.17)

6. Disproportionation
causes
t e
disappearance
of
two
molecules
of
alkylate
to
give
a

lower
and
a
ig er
molecular
weig t
isoparaffin
t an
t e
initial
one:

2iC8H18

→
iC7H16

+
C9H20















(2.18)

7.

Larger
isoalkyl
cations
can
crack,
leading
to
smaller
isoalkyl
cations
and
olefins:

+ →
iC5H11

+
iC7H14














(2.19)
















































+



















iC12H25































































+








































→



iC6H13

+
iC6H12













(2.20)


+ +






iC6H13

→
iC5H11

+
iC5H10

+
iC6H12














(2.21)





















































8. Self!alkylation
accounts
for
t e
formation
of
trimet yl!pentanes
w en
isobutane
is

alkylated
wit
olefins
ot er
t an
butanes.
At
t e
same
time,
saturated
paraffin
of
t e

same
carbon
number
as
t e
olefin
is
obtained.
T e
reaction
sc eme
for
pentene,
for

example
is
:

+











+
C
!
C
!
C
!
C

+
C
!
C
!
C




C
!
C
!
C
!
C

+

C
!
C
!
C












(2.22)








|













|








|





|








C



C







C




C



+
C
!
C
!
C




C
!
C
=
C

+
H+
















(2.23)


|











|






C










C

29

C



+











|









+

C
!
C
!
C
+
C
!
C
=
C

→

C
!
C
!
C
!
C
!
C





























(2.24)









|



















|










|









|



C










C








C








C


T e
above
reactions
are
believed
to
be
fundamental
to
t e
alkylation
process
and
are
used
to
explain
t e
formation
of
bot
primary
and
secondary
products.
Alt oug
isobutane
and
butenes
were
used
as
examples,
t ese
reactions
also
applied
to
ot er
isoparaffins
and
olefins.

2.2.4 eaction Mechanism for Alkylation of Isobutane with Propylene

Langley
and
Pike
(1972)
studied
t e
sulfuric
acid
alkylation
of
isobutane
wit
propylene
and
proposed
seventeen!reaction
mec anism
model
(Table
2.2)
based
on

Sc mering
carbonium
ion
mec anism
wit
modification
introduced
to
account
for
iC9
and
iC10
formation.
Experimental
measurements
were
made
in
an
ideally
mixed,
continuous
flow
stirred
tank
reactor
at
t e
temperature
range
18!57
oC.
T e
model
was
found
to
be
valid
in
t e
range
of
27!57
oC
using
95%
sulfuric
acid
catalyst.
Lower
concentration
of
sulfuric
acid
(around
90%)
resulted
in
increased
rates
of
formation
of
iC9
and
iC10
and
decreased
rates
of
alkylate
formation.

2.2.5 eaction Mechanism for Alkylation of Isobutane with Butylene and

Pentylene

Vic ailak
(1995)
extended
t e
mec anism
given
by
Langley
and
Pike
(1972)
to
a
nineteen!reaction
mec anism
model
for
isobutane
wit
butylene
alkylation
and
twenty
one!reaction
mec anism
model
for
isobutane
wit
pentylene
alkylation.
T e
reaction
mec anism
model
for
alkylation
of
isobutane
wit
butylene
is
s own
in
Table
2.3.
T e
reaction
rate
constants
(frequency
factor
and
activation
energy)
for
eac
reaction
of

30

t ese
mec anisms
t at
were
taken
to
be
identical
to
t e
reactions
in
t e
propylene
mec anism.
A
list
of
t ese
rate
constants,
k
values,
are
given
in
Appendix
C.8.
T e
results
from
t is
reaction
model
s owed
a
fair
agreement
(90%)
wit
t e
publis ed
data

(Albrig t,
1992,
Sric anac aikul,
1996).
T e
reaction
mec anism
and
rate
constants
s own
in
Table
2.3
to
describe
t e
alkylation
reaction
were
used
for
t e
STRATCO
reactors
in
t e
Motiva
plant
model.

31

Table
2.2.
Reaction
Mec anism
and
Material
Balances
for
Sulfuric
Acid
Alkylation
of
Isobutane
wit
Propylene
(Langley,
1969)


Initiation
reactions




(a)
Material
balance
on
reactants
and














Associated
consumption
rates.

+ + = k1 + − − − 3 + → 3 − = + + C HX C X



r C4 k 2[C 3 X ][ C4 ] k 3[ C 5 X ][ C4 ]

+ k + + + − + →2 + − − + − + C3 X C 4 C 3 C 4 X




k 4[ C 6 X ][ C 4] k 5[ C 7 X ][ C 4 ]
+ − + + − +







k 6[ C 8 X ][ C 4] k 7[ C 9 X ][ C 4 ]

+ − Primary
reactions






k 8[ C 10 X ][ C 4 ]

+ = k + = + = − + 11 → − − = + − + C4 X C3 C7 X


rC 3= k1[C 3 ][HX ] k11 [ C 4 X ][C3 ]

+ k + + = − + →5 + − − C 7 X C 4 C 7 C 4 X






k 15[ C 7 X ][C 3 ]
Self-alkylation reactions

(b)
Product
formation
equations.

+ k = + − →9 + = − C 4 X C 4 HX




rC3 k 2[CX 3 ][ C 4 ]

+ = k + + − + 10 → − = − C 4 X C 4 C 8 X


r C 5 k 3[ C 5 X ][ C 4

+ k + + − + →6 + − = − C 8 X C 4 C 8 C 4 X


r C 6 k 4[ C 6 X ][ C 4 ]
= + −





r C 7 k 5[ C 7 X ][ C 4 ]
= + − Destructive
alkylation
reactions



r C8 k 6[ C 8 X ][ C 4 ]

+ k = + − 12 → + = − C 7 X C 7 HX


r C 9 k 7[ C 9 X ][ C 4 ]

= + k = + + + − 13 → + − = − C 7 C 4 X C 5 C 6 X

r C10 k 8[ C 10 X ][ C 4 ]

= k + + 14 → − C5 HX C5 X



(c)
Olefinic
intermediate
rate
equations.

k + + = + + − + →3 + − = = − − − C 5 X C 4 C 5 C 4 X

r C 4 = 0 k 9[ C 4 X ] k10 [ C 4][ C 4 X ]

+ k + = + + − + →4 + − = = − + − − C6 X C4 C 6 C4 X


r C 5 = 0 k13[ C7 ][ C4 X ] k17 [ C10 X ]

+ = k + = = + − + 15 → − − − C 7 X C 3 C10 X




























k 14[ C 5 ][HX ] k16 [ C 5][ C 4 X ]

+ k + + = + − + →8 + − = = − − − C10 X C 4 C10 C 4 X

r C 7 = 0 k12 [ C 7 X ] k13 [ C 7 ][ C 4 X ]

= + k + + − 16 → − C 5 C 4 X C 9 X


(d)
Carbonium
ion
rate
equations.

+ k + = + − + →7 + − = = − − C 9 X C 4 C 9 C 4 X


rC 3+ X − 0 k1[C 3 ][HX ] k 2[C 3 X ][ C 4 ]

+ k = + + = + − 17 → + − = = − − − − − C10 X C 5 C 5 X


r C 4 + X − 0 r C 4 k 9[ C 4 X ] k10 [ C 4 ][ C 4 X ]
+ − = − = + − −






k 11[ C 4 X ][C 3 ] k13 [ C 7][ C 4 X ]

= + −











k 16[ C 5 ][ C 4 X ]
= = = + + − −





r C 5 + X − 0 k14 [ C 5 ][HX ] k17 [ C 10 X ]

+ −






k 3[ C 5 X ][ C 4 ]
= = = + − − + −





r C 6 + X − 0 k13 [ C 7 ][ C X ] k 4[ C 6 X ][ C 4 ]
= = + − = − + − −





r C 7 + X − 0 k11 [ C 4 X ][C 3] k 5[ C 7 X ][ C 4 ]
+ − = − + −






k 15[ C 7 X ][C 3 ] k 12[ C 7 X ]
= = = + − − + −





r C8 + X − 0 k10 [ C 4][ C 4 X ] k 6[ C 8 X ][ C 4 ]
= = = + − − + −





r C 9 + X − 0 k16 [ C 5][ C 4 X ] k 7 [ C 9 X [ C 4 ]
= = + − = − + − −





r C10 + X − 0 k15 [ C 7 X ][C 3 ] k17 [ C 10 X ]

+ −








32
k 8[ C 10 X ][ C 4 ]

Table
2.3.
Reaction
Mec anism
and
Material
Balances
for
Sulfuric
Acid
Alkylation
of
Isobutane
wit
Butylene
(Vic ailak,
1995)

Initiation
reactions




(a)
Material
balance
on
reactants
and














Associated
consumption
rates.

= + k1 + − − = + − + + − + C 4 HX → C4 X


r C4 k 2[CX 4 ][ C4] k 3[ C 5 X ][ C4 ]

+ − k + − + − + − + →2 + + + C4 X C 4 C 4 C 4 X

















k 4[ C 6 X ][ C 4] k 5[ C 7 X ][ C 4 ]
+ − + + − +




















k 6[ C 8 X ][ C 4] k 7[ C 9 X ][ C 4 ]
+ − + − Primary
reactions



















k8[ C10 X ][ C 4 ] + k18[ C11 X ][ C 4 ]

+ − = k + − = + − = + →11 − = + + C 4 X C 4 C 8 X


rC 4= k1[C 4 ][HX ] k11 [ C 4 X ][C 4 ]

+ − k + − + − = + − = + →6 + C 8 X C 4 C 8 C 4 X


















k 15[ C 7 X ][C 4] + k 19[ C 6 X ][C 4 ]

Self!alkylation
reactions



(b)
Product
formation
equations.

+ k = + − →9 + = − C 4 X C 4 HX




rC 4 k 2[CX 4 ][ C 4 ]

+ = k + + − + 10 → − = − C 4 X C 4 C 8 X


r C 5 k 3[ C 5 X ][ C 4 ]

+ k + + − + →6 + − = − C 8 X C 4 C 8 C 4 X


r C 6 k 4[ C 6 X ][ C 4 ]
= + −





r C 7 k 5[ C 7 X ][ C 4 ]
= + −





r C8 k 6[ C 8 X ][ C 4 ]
= + − Destructive
alkylation
reactions



r C 9 k7 [ C9 X ][ C 4 ]

+ − k = + − 12 → + = C8 X C8 HX



r C 10 k8[ C10 X ][ C4 ]

= + k = + + − 13 → + − = + − C8 C 4 X C5 C 7 X


r C11 k18 [ C 11 X ][ C 4 ]

= k + + 14 → − C5 HX C5 X



(c)
Olefinic
intermediate
rate
equations

+ − k + − + = + + →3 + = = − − − C 5 X C 4 C 5 C 4 X


r C 4 = 0 k 9[ C 4 X ] k10 [ C 4][ C 4 X ]

+ − k + − = + + + →5 + = = − + − − C 7 X C 4 C 7 C 4 X


r C 5= 0 k13 [ C 8 ][ C 4 X ] k17 [ C 11 X ]

+ − = k + − = = + − + →15 − C 7 X C 4 C11 X




























k 14[ C 5 ][HX ] k16 [ C 5][ C 4 X ]

+ − k + − + − = + − + →18 + = = − C11 X C 4 C11 C 4 X


r C8= 0 k12 [ C 8 X ] k13 [ C 8 ][ C 4 X ]

= + k + + − 16 → − C 5 C 4 X C 9 X


(d)
Carbonium
ion
rate
equations.

+ k + = + − + →7 + − = = − − C 9 X C 4 C 9 C 4 X


rC 4+ X − 0 k1[C 4 ][HX ] k 2[C 4 X ][ C 4 ]

+ − k = + − + − = + − →17 + = = − − − − C11 X C 5 C 6 X















r C 4+ X − 0 r C 4 k9[ C 4 X ] k10 [ C 4 ][ C 4 X ]
+ − = − = + − − = + −












k 11[ C 4 X ][C 4] k 13[ C 8][ C 4 X ] k 16[ C 5][ C 4 X ]
= = = − + −


r C5+ X − 0 k14 [ C 5 ][HX ] k 3[ C 5 X ][ C 4 ]










+ + = = − − − − + − =


r C 6+ X − 0 k17 [ C11 X ] k 4[ C 6 X ][ C 4 ] k19 [ C 6 X ][C 4 ]

+ − = + − + = + − + →k 4 + + − = = − − − C6 X C 4 C6 C 4 X


r C 7+ X − 0 k13 [ C 4 X ][ C8 ] k 5[ C 7 X ][ C 4 ] k 15 [ C 7 X ][C4 ]

= + + + − + = →k 19 + − = = − + − = − C 6 X C 4 C10 X


r C8+ X − 0 k11 [C 4][ C 4 X ] k10 [ C 4 X ][ C 4 ]

+ − + →k 8 + + − + − − + − C10 X C 4 C10 C 4 X










k 6[ C 8 X ][ C 4] k 12[ C 8 X ]
= = = + − − + −





r C 9 + X − 0 k 16[ C 5][ C 4 X ] k 7 [ C 9 X [ C 4 ]
= = + − = − + −





r C10+ X − 0 k19 [ C 6 X ][C 4] k 8[ C 10 X ][ C 4 ]
= = + − = − + − − + −





r C11+ X − 033
k15 [ C 7 X ][C 4 ] k18 [ C 11 X ][ C 4 ] k 17[ C 11 X ]

2.2.6 Influence of Process Variables

T e
most
important
process
variables
are
reaction
temperature,
acid
strengt ,
isobutane
concentration,
and
olefin
space
velocity.
C anges
in
t ese
variables
affect
bot
product
quality
and
yield
(Gary
and
Handwerk,
1984).

Reaction
Temperature

T e
reaction
temperature
in
sulfuric
acid
alkylation
is
usually
in
t e
range
of
32!

50
oF.
At
ig er
temperatures,
oxidation
reactions
become
important
and
acid
consumption
increases
(Corma
and
Martinez,
1993).
At
temperatures
above
65
oF,
polymerization
of
t e
olefins
becomes
significant
and
yields
are
decreased.
If
t e
operation
takes
place
at
lower
temperatures,
t e
effectiveness
decreases
due
to
t e
increase
in
t e
acid
viscosity
and
t e
decreased
solubility
of
ydrocarbons
in
t e
acid
p ase
(Gary
and
Handwerk,
1984).

Acid
Strengt

Acid
strengt
as
varying
effects
on
alkylate
quality,
depending
on
t e
effectiveness
of
t e
reactor
mixing
and
t e
water
content
of
t e
acid.
In
sulfuric
acid
alkylation,
t e
best
quality
and
ig est
yields
are
obtained
wit
acid
strengt s
of
93!

95%
by
weig t
of
acid,
1!2%
of
water
and
t e
remainder
ydrocarbon
diluents.
T e
water
content
in
t e
acid
lowers
its
catalytic
activity
by
about
3!5
times
as
muc
as

ydrocarbon
diluents,
t us,
an
88%
acid
containing
5%
water
is
muc
less
effective
catalyst
t an
t e
same
strengt
acid
containing
2%
water.
At
concentrations
ig er
t an

99%,
isobutane
reacts
wit
SO3,
and
below
85!88%
concentration,
t e
catalyst
becomes
a
polymerization
rat er
t an
an
alkylation
catalyst.
Poor
mixing
in
a
reactor
requires

ig er
acid
strengt
necessary
to
keep
acid
dilution
down.
Increasing
t e
acid
strengt

34

from
89%
to
93%
increases
t e
alkylate
quality
by
1!2
octane
numbers
(Cupit
et.al.,

1962).

Isobutane
Concentration

Isobutane
concentration
in
t e
feed
to
t e
reactor
is
generally
expressed
in
terms
of
isobutane/olefin
ratio.
T is
is
one
of
t e
most
important
process
variables
t at
controls
acid
consumption,
yield
and
quality
of
t e
alkylate.
W en
t e
isobutane/olefin
ratio
in
t e
feed
is
ig
(15:1),
t e
olefin
is
more
likely
to
react
wit
an
isobutane
molecule
to
form
t e
desired
product
t an
to
undergo
butene!butene
polymerization.

T us
undesired
reactions
are
minimized.
If
t is
ratio
is
kept
low
(<5:1),
production
of

eavy
polymers
increases
due
to
polymerization
reactions,
and
t e
acid
consumption
increases
(Corma
and
Martinez,
1993).
An
increase
in
t is
ratio
produces
a
decrease
in
acid
consumption
w ile
increasing
bot
t e
yield
and
t e
quality
of
t e
alkylate.
T e
external
isobutane
to
olefin
ratio
in
sulfuric
acid
plants
are
usually
in
t e
range
of
5:1
to

15:1
(Gary
and
Handwerk,
1984).

Olefin
Space
Velocity

Olefin
space
velocity
is
defined
as
t e
volume
of
olefin
c arged
per
our
divided
by
t e
volume
of
acid
in
t e
reactor.

Volumetr c Flow Rate of Olef n
Olef n space veloc ty =
Vol. Fract on of ac d n reactor *Volume of reactor

Lowering
t e
olefin
space
velocity
reduces
t e
amount
of
ig
boiling
ydrocarbons
produced,
increase
t e
product
octane
and
lowers
acid
consumption
(Gary
and

Handwerk,
1984).
Olefins
space
velocity
is
one
way
of
expressing
space!time,
anot er
is
by
using
contact
time.

35

2.2.7 Feedstock

Olefins
and
isobutane
are
used
as
alkylation
feedstocks.
T e
c ief
sources
of
olefins
are
catalytic
cracking
and
coking
operations.
Butenes
and
propenes
are
t e
most
common
olefins
used
but
et ylene
and
pentenes
are
included
in
some
cases.
Olefins
can
be
produced
by
de ydrogenation
of
paraffins
and
isobutane
is
cracked
commercially
to
provide
alkylation
unit
feed.

Hydrocrackers
and
catalytic
crackers
produce
a
majority
of
t e
isobutane
used
in
alkylation,
moreover
it
is
obtained
from
catalytic
reformers,
crude
distillation,
and
natural
gas
processing.
In
some
cases,
normal
butane
is
isomerized
to
produce
additional
isobutane
for
alkylation
unit
feed.

2.2.8 Products

In
addition
to
t e
alkylate
stream,
t e
products
leaving
t e
alkylation
unit
include
t e
propanes
and
normal
butane
t at
enter
wit
t e
saturated
and
unsaturated
feed
streams
as
well
as
a
small
quantity
of
tar
produced
by
polymerization
reactions.

T e
product
streams
leaving
an
alkylation
unit
are:

1. LPG
grade
propane
liquid

2. Normal
butane

liquid

+ 3. C5
Alkylate

4. Spent
Acid
(wit
tar)

Only
about
0.1
%
by
volume
of
olefin
feed
is
converted
into
tar.
T is
is
not
truly
a
tar
but
a
t ick
dark
brown
oil
containing
complex
mixtures
of
conjugated
cyclopentadienes
wit
side
c ains
(T omas,
1970).

36

2.2.9 Catalysts

Concentrated
sulfuric
and
ydroflouric
acids
are
t e
only
catalysts
used
commercially
today
for
t e
production
of
ig
octane
alkylate
gasoline
but
ot er
catalysts
are
used
to
produce
et ylbenzene,
cumene
and
long
c ain
(C12
to
C16)
alkylated
benzenes
(T omas,
1970).

T e
desirable
reactions
are
t e
formation
of
C8
carbonium
ions
and
t e
subsequent
formation
of
alkylates.
T e
main
undesirable
reaction
is
polymerization
of
olefins.
Only
strong
acids
can
catalyze
t e
alkylation
reaction
but
weaker
acids
can
cause
polymerization
to
take
place.
T erefore
t e
acid
strengt s
must
be
kept
above
88

%
by
weig t
H2SO4
or
HF
in
order
to
prevent
excessive
polymerization.
Sulfuric
acid
containing
free
SO3
also
causes
undesired
side
reactions
and
concentrations
greater
t an

99.3
%
H2SO4
are
not
generally
used
(T omas,
1970).

Isobutane
is
soluble
in
t e
acid
p ase
only
to
t e
extent
of
about
0.1
%
by
weig t
in
sulfuric
acid
and
about
3
%
in
ydroflouric
acid.
Olefins
are
more
soluble
in
t e
acid
p ase
and
a
slig t
amount
of
polymerization
of
t e
olefins
is
desirable
as
t e
polymerization
products
dissolve
in
t e
acid
and
increase
t e
solubility
of
isobutane
in
t e
acid
p ase.

If
t e
concentration
of
t e
acid
becomes
less
t an
88
%,
some
of
t e
acid
must
be
removed
and
replaced
wit
stronger
acid.
In
ydroflouric
acid
units,
t e
acid
removed
is
redistilled
and
t e
polymerization
products
removed
as
t ick
dark
oil.
T e
concentrated
HF
is
recycled
in
t e
unit
and
t e
net
consumption
is
about
0.3
lb
per
barrel
of
alkylate
produced
(Templeton
and
King,
1956).

37

T e
sulfuric
acid
removed
must
be
regenerated
in
a
sulfuric
acid
plant
w ic
is
generally
not
part
of
t e
alkylation
unit,
and
t e
acid
consumption
ranges
from
18
to
30
lb
per
barrel
of
alkylate
produced.
Makeup
acid
is
usually
99.3
%
by
weig t
H2SO4.

2.3 Summary

T is
c apter
reviewed
t e
literature
for
c emical
process
analysis
and
for
t e
current
understanding
of
alkylation
process.
T e
next
c apter
describes
t e
met odology
of
Advanced
Process
Analysis
System.
Subsequent
c apters
describe

Motiva’s
Alkylation
process
and
t e
results
of
applying
t e
system
to
t is
process.

38

CHAPTE 3 METHODOLOGY In t is c apter a detailed description is given for t e met odology used in t e

Advanced Process Analysis System. T e framework for t e Advanced Process Analysis

System was s own in Figure 1.1. T e main components of t is system are a flows eeting program for process material and energy balances, an on-line optimization program, a c emical reactor analysis program, a eat exc anger network design program and a pollution assessment module. An overview of eac of t ese programs was given in C apter 1.

C emical Reactor

Separation and Recycle

Heat Exc anger Network

Utilities

Figure 3.1: ‘Onion Skin’ Diagram for Organization of a C emical Process and Hierarc y of Analysis.

T e Advanced Process Analysis System met odology to identify and eliminate t e causes of energy inefficiency and pollutant generation is based on t e onion skin diagram s own in Figure 3.1. Having an accurate description of t e process from online optimization, an evaluation of t e best types of c emical reactors is done first to modify and improve t e process. T en t e separation units are evaluated. T is is followed by t e pinc analysis to determine t e best configuration for t e eat exc anger network and determine t e utilities needed for t e process. Not s own in t e diagram is t e

39

pollution index evaluation, w ic is used to identify and minimize emissions. T e following gives a detailed description of t e components of t e Advanced Process

Analysis System and ow t ey are used toget er to control and modify t e process to maximize profit and minimize wastes and emissions.

3.1 Flowsheeting Program

T e first step towards implementing t e Advanced Process Analysis System is t e development of t e process model, w ic is also known as flows eeting. T e process model is a set of constraint equations, w ic represent a mat ematical model of material and energy balances, rate equations and equilibrium relation for t e process.

Formulation of t e process model can be divided into two important steps.

Formulation of Constraints for Process Units: A process model can be formulated eit er empirically or mec anistically. A mec anistic model makes use of constraint equations depicting conservation laws (mass and energy balances), equilibrium relations and empirical formulas. Advanced Process Analysis System uses mec anistic models for analysis.

Mat ematically, constraints fall into two types: equality constraints and inequality constraints. Equality constraints are material and energy balances or any ot er exact relations ip in a process. Inequality constraints include demand for product, availability of raw materials and capacities of process units.

Classification of Variables and Determination of Parameters: After t e constraints are formulated, t e variables in t e process are divided into two groups: measured and unmeasured variables. Measured variables are t e variables w ic are directly measured from t e distributed control systems (DCS) and t e plant control

40

laboratory. T e remaining variables are t e unmeasured variables. For redundancy, t ere must be more measured variables t an t e degree of freedom of t e equality constraints.

Parameters in t e model can also be divided into two types: constant and time varying parameters. Constant parameters do not c ange wit time and include reaction activation energy, eat exc anger areas. Time-varying parameters include fouling factors in eat exc angers and catalyst deactivation parameters. T ey c ange slowly wit time and are related to t e degradation of performance of equipment.

Flowsim: T e program used for flows eeting in t e Advanced Process

Analysis System is called ‘Flowsim’. Flowsim provides a grap ical user interface wit interactive capabilities. An example of t is interface is s own in Figure 3.2 for a simple refinery, and t e comparable diagram for t e alkylation process is given in t e next c apter, Figure 4.5. Process units are represented as rectangular s apes w ereas t e process streams are represented as lines wit arrows between t ese units. Process units and streams can be drawn by clicking t e corresponding icons on t e Flowsim window.

Eac process unit and stream in t e flows eet must ave a name and a description.

Process information is divided into t e following six categories; equality constraints, inequality constraints, unmeasured variables, measured variables, parameters and constants.

T e information in t e first five categories is furt er classified by associating it wit eit er a unit or a stream in t e flows eet. For example, for a unit t at is a eat exc anger, t e relevant information includes t e mass balance and eat transfer

41

equations, limitations on t e flowrates and temperatures if any, t e eat transfer coefficient parameter and all t e intermediate variables defined for t at exc anger.

Figure 3.2 Example of Flowsim Screen for a Simple Refinery

For a stream, t e information includes its temperature, pressure, total flowrate, molar flowrates of individual components etc. Information not linked to any one unit or stream is called t e ‘Global Data’. For example, t e overall daily profit of t e process is a global unmeasured variable because it is not related to any particular process unit or stream.

42

T e sixt category of constants can be grouped into different sets based on t eir p ysical significance. For example, constants related to eat exc angers can be placed in one group and t ose related to reactors into anot er group.

Flowsim also as a sevent category of information called as t e ‘ent alpy coefficients’. T is stores t e list of all t e c emical components in t e process and t eir ent alpy coefficients for multiple temperature ranges. All of t is process information is entered wit t e elp of t e interactive, user-customized grap ic screens of Flowsim.

T is concludes t e description of t e flows eeting part of t e Advanced Process

Analysis System. Appendix J is t e user’s manual, w ic gives a step-by-step description of creating process models.

3.2 Online Optimization Program

Once t e process model as been developed using Flowsim, t e next step is to conduct on-line optimization. On-line optimization is t e use of an automated system w ic adjusts t e operation of a plant based on product sc eduling and production control to maximize profit and minimize emissions by providing setpoints to t e distributed control system. As s own in Figure 1.5, it includes t ree important steps: combined gross error detection and data reconciliation, simultaneous data reconciliation and parameter estimation and plant economic optimization. In combined gross error detection and data reconciliation, a set of accurate plant measurements is generated from data extracted from t e plant’s Distributed Control System (DCS). T is set of data is used for estimating t e parameters in plant models. Parameter estimation is necessary to ave t e plant model matc t e current performance of t e plant. T en economic optimization is conducted to optimize t e economic model using t is current plant

43

model as constraints. T is generates set points for t e distributed control system to move t e plant to optimal operating conditions.

Eac of t e above t ree-optimization problems in on-line optimization as a similar mat ematical statement as following:

Optimi e: Objective function Subject to: Constraints from plant model. w ere t e objective function is a joint distribution function for data reconciliation, least squares for parameter estimation and a profit function (economic model) for plant economic optimization. T e constraint equations describe t e relations ip among variables and parameters in t e process, and t ey are material and energy balances, c emical reaction rates, t ermodynamic equilibrium relations, and ot ers.

To perform data reconciliation, t ere as to be redundancy in t e measurements, i.e. t ere s ould be more measurements t an t e degrees of freedom in t e process model. For redundancy, t e number of measurements to determine t e minimum number of measured variables is given by t e degree of freedom, w ic is calculated using t e following equation (Felder and Rousseau, 1986).

Degree of freedom = Total number of variables – Total number of equality

constraints + Number of independent chemical reactions.

Also, t e unmeasured variables ave to be determined by t e measured variables, called observability. If an unmeasured variable can not be determined by a measured variable, it is unobservable. T is is called t e ‘observability and redundancy criterion’, w ic needs to be satisfied (C en, 1998).

Combined Gross Error Detection and Data econciliation: Process data from distributed control system is subject to two types of errors, random errors and

44

gross errors. Gross errors must be detected and rectified before t e data is used to estimate plant parameters. Combined gross error detection and data reconciliation algorit ms can be used to detect and rectify gross errors in measurements for on-line optimization. T ese algorit ms are: measurement test met od using normal distribution,

Tjoa-Biegler’s met od using contaminated Gaussian distribution, and robust statistical met od using robust functions. T e t eoretical performance of t ese algorit ms as been evaluated by C en, 1998.

Based on C en’s study, Tjao-Biegler’s met od or robust met od is used to perform combined gross error detection and data reconciliation. It detects and rectifies gross errors in plant data sampled from distributed control system. T is step generates a set of measurements containing only random errors for parameter estimation. T en, t is set of measurements is used for simultaneous parameter estimation and data reconciliation using t e least-squares met od.

Simultaneous Data econciliation and Parameter Estimation: T e general met odology for t is step is similar to t e met odology of combined gross error detection and data reconciliation. T e difference is t at parameters in plant model are considered as variables along wit process variables in simultaneous data reconciliation and parameter estimation rat er t an being constants as in data reconciliation. Bot process variables and parameters are simultaneously estimated. Based on C en’s study, t e least squares algorit m is used to carry out combined gross error detection and data reconciliation. T e data set produced by parameter estimation is free of any gross errors, and t e updated values of parameters represent t e current state of t e process.

T ese parameter values are used in economic optimization.

45

Plant Economic Optimization: T e objective of plant economic optimization is to generate a set of optimal operating setpoints for t e distributed control system. T is set of optimal setpoints will maximize t e plant profit, satisfy t e current constraints in plant model, meet t e requirements for t e demand of t e product and availability of raw materials, and meet t e restriction on pollutant emission. Optimization can be ac ieved by maximizing t e economic model (objective function) subject to t e process constraints. T e objective function can be different depending on t e goals of optimization. T e objectives can be to maximize plant profit, minimize energy use, minimize undesired by-products, minimize waste/pollutant emission or a combination of t ese objectives. T e result of economic optimization is a set of optimal values for all t e measured and unmeasured variables in t e process. T ese are t en sent to t e distributed control system (DCS) to provide setpoints for t e controllers.

On-line optimization program of Advanced Process Analysis System retrieves t e process model and t e flows eet diagram from Flowsim. Additional information needed to run online optimization includes plant data and standard deviation for measured variables; initial estimates, bounds and scaling factors for bot measured and unmeasured variables; and economic objective function. T e program t en constructs t ree optimization problems s own in Figure 1.5 and uses GAMS (General Algebraic

Modeling System) to solve t em sequentially. Results of t ese t ree problems can be viewed using t e grap ical interface of Flowsim. T is is illustrated in t e user’s manual in Appendix J.

46

3.3 Chemical eactor Analysis Program

Having optimized t e process operating conditions for t e most current state of t e plant, t e next step in Advanced Process Analysis System is to evaluate modifications to improve t e process and reduce emission and energy consumption.

First, c emical reactors in t e process are examined. T e reactors are t e key units of c emical plants. T e performance of reactors significantly affects t e economic and environmental aspects of plant operation. T e formulation of constraints in t ese types of units is very important and complicated owing to t e various types of reactors and complex reaction kinetics. Unlike a eat exc anger w ose constraints are similar regardless of types of equipment, t ere is a great variation in deriving t e constraints for reactors.

T e c emical reactor analysis program of Advanced Process Analysis System is a compre ensive, interactive computer simulation t at can be used for modeling various types of reactors suc as plug flow, CSTR and batc reactors. T is is s own in Figure

1.3. Reaction p ases included are omogeneous gas, omogeneous liquid, catalytic liquid, gas-liquid etc. T e options for energy models include isot ermal, adiabatic and non-adiabatic.

T e kinetic data needed for t e reactor system includes t e number of reactions taking place in t e reactor and t e number of c emical species involved. For eac reaction, stoic iometry and reaction rate expressions also need to be supplied. P ysical properties for c emical species can be retrieved from Flowsim.

Feed stream for t e reactor is obtained from Flowsim and its temperature, pressure and flowrate are retrieved using t e results from on-line optimization. Finally,

47

t e dimensions of t e reactor and eat transfer coefficients are supplied. All of t is data is used to simulate reactor conditions and predict its performance. Reactant concentration, conversion, temperature and pressure are calculated as functions of reactor lengt or space-time. T e results can be viewed in bot tabular and grap ical form. T e user’s manual in Appendix J gives a step-by-step description of t e program.

As operating process conditions c ange, t e performance of reactors can vary to a significant extent, also. T e reactor design program a wide range of different types of reactors, w ic can be examined and compared to decide t e best reactor configuration for economic benefits and waste reduction.

3.4 Heat Exchanger Network Program

T e optimization of c emical reactors is followed by eat exc anger network optimization as s own in t e onion skin diagram in Figure 3.1. Most c emical processes require eating and cooling of certain process streams before t ey enter anot er process unit or are released into environment. T is eating or cooling requirement can be satisfied by matc ing t ese streams wit one anot er and by supplying external source of eating or cooling. T ese external sources are called as utilities, and t ey add to t e operating cost of t e plant. T e Heat Exc anger Network program aims at minimizing t e use of t ese external utilities by increasing energy recovery wit in t e process. It also synt esizes a eat exc anger network t at is feasible and as a low investment cost.

T ere are several ways of carrying out t e above optimization. Two of t e most important ones are pinc analysis and mat ematical programming met ods. Pinc analysis is based on t ermodynamic principles w ereas t e mat ematical met ods are

48

based on mass and energy balance constraints. T e Heat Exc anger Network Program

(abbreviated as THEN) is based on t e met od of pinc analysis.

T e first step in t e implementation of THEN is t e identification of all t e process streams, w ic are important for energy integration. T ese streams usually include streams entering or leaving eat exc angers, eaters and coolers. T e flows eeting diagram of Flowsim is used in t e selection of t ese streams.

T e next step in t is optimization task involves retrieval of necessary information related to t ese streams. Data necessary to perform eat exc anger network optimization includes temperature, flowrate, film eat transfer coefficient and ent alpy data. Ent alpy data can be in t e form of constant eat capacities for streams wit small temperature variations. For streams wit large variations, it can be entered as temperature-dependent ent alpy coefficients. Film eat transfer coefficients are needed only to calculate t e areas of eat exc angers in t e new network proposed by THEN.

Temperature and flowrates of various process streams are automatically retrieved from t e results of online optimization. T e setpoints obtained after t e plant economic optimization are used as t e source data. P ysical properties suc as t e eat capacities, ent alpy coefficients and film eat transfer coefficients are retrieved from

Flowsim.

T e t ird step in eat exc anger network optimization is t e classification of streams into ot streams and cold streams. A ot stream is a stream t at needs to be cooled to a lower temperature w ereas a cold stream is a stream t at needs to be eated to a ig er temperature. Usually, streams entering a cooler or t e ot side of a eat exc anger are ot streams w ereas streams entering t roug a eater or t e cold side of

49

a eat exc anger are cold streams. T e final step in t is problem requires t e specification of minimum approac temperature. T is value is usually based on experience.

Having completed all of t e above four steps, eat exc anger network optimization is now performed using THEN. T ermodynamic principles are applied to determine t e minimum amount of external supply of ot and cold utilities. Grand

Composite Curve is constructed for t e process, w ic s ows eat flows at various temperature levels. A new network of eat exc angers, eaters and coolers is proposed, w ic features t e minimum amount of external utilities. T is network drawn in a grap ical format is called t e Network Grid Diagram. An example of a network grid diagram is given in Figure 1.4. Detailed information about t e network can be viewed using t e interactive features of t e user interface. T e user’s manual in Appendix J provides a step-by-step description of t e program.

T e amount for minimum ot and cold utilities calculated by t e Heat

Exc anger Network Program is compared wit current amount of utilities being used in t e process. If t e current amounts are greater t an t e minimum amounts, t e process

as potential for reduction in operating cost. T e network grid diagram synt esized by

THEN can be used to construct a eat exc anger network t at ac ieves t e target of minimum utilities. Savings in operating costs are compared wit cost of modification of t e existing network, and a decision is made about t e implementation of t e solution proposed by THEN.

50

3.5 Pollution Assessment Program

T e final step in t e Advanced Process Analysis System is t e determination of t e emissions from t e process and t e location w ere t ese pollutants are generated wit in t e process. T e pollution assessment module of Advanced Process Analysis

System is called ‘T e Pollution Index Program’. It is based on t e Waste Reduction

Algorit m and t e Environmental Impact T eory as described in C apter 2. It defines a quantity called as t e pollution index to provide a basis for measuring t e pollution generated by t e process.

Environmental impact of a c emical process is caused by t e streams t at t e process takes from and emits to t e environment. Only t ese input and output streams are considered in performing pollution index analysis. Ot er streams, w ic are completely internal to t e process, are excluded. In t e Pollution Index Program, t is selection of input-output streams is automatically done based on t e plant information entered in Flowsim.

T e next step in pollution index analysis is t e classification of output streams into product and non-product streams. All streams w ic are eit er sold as product or w ic are used up in a subsequent process in t e production facility are considered as product streams. All ot er output streams, w ic are released into t e environment, are considered as non-product streams. All non-product streams are considered as pollutant streams w ereas all product streams are considered to ave zero environmental impact.

Pollution index of a stream is a function of its composition. T e composition data for t e streams is retrieved from t e results of on-line optimization performed earlier. T is can be eit er in terms of molar flowrates or fractions. Additional data

51

needed for pollution analysis is t e specific environmental impact potential values of individual c emical species present in a stream. T ese values for various c emical species are available in t e report on environmental life cycle assessment of products.

T e last piece of information required is t e relative weig ting factors for t e process plant. According to t e Waste Reduction Algorit m, environmental impact of c emical processes can be broadly classified into nine categories e.g. acidification, p otoc emical oxidation etc. T e relative weig ting factors allow t e customization of t e analysis to local conditions. T eir values depend on t e location of t e plant and its surrounding conditions. For example, t e weig ting factor for p otoc emical oxidation is ig er in areas w ic suffer from smog.

Having finis ed all of t e above prerequisite steps, t e pollution index program is now called to perform t e analysis. Mass balance constraints are solved for t e process streams involved, and equations of t e Environmental Impact T eory are used to calculate t e pollution index values. Six types of pollution indices are reported for t e process. T ree of t ese are based on internal environmental efficiency w ereas t e ot er t ree are based on external environmental efficiency. Hig er t e values of t ese indices,

ig er is t e environmental impact of t e process.

T e pollution index program also calculates pollution indices for eac of t e individual process streams. T ese values elp in identification of t e streams w ic contribute more to t e overall pollution impact of t e process. Suitable process modifications can be done to reduce t e pollutant content of t ese streams.

Every run of on-line optimization for t e process can be followed by pollution index calculations. T e new pollution index values are compared wit t e older values.

52

T e comparison s ows ow t e c ange in process conditions affects t e environmental impact. T us, t e pollution index program can be used in continuous on-line monitoring of t e process.

3.6 Summary

To summarize, t e Flowsim program is used first to develop t e process model.

T e On-Line Optimization program is used to control t e process to maximize profit and minimize pollutants, and in doing t is it as to continually update t e process model to matc t e current state of t e plant. T is provides an accurate description of t e process for off-line process modification analysis. T e C emical Reactor Analysis program is used to determine t e best type of c emical reactors for t e process and to optimize t eir performance. T e Heat Exc anger Network Program t en optimizes t e

eat recovery in t e system and minimizes t e amount of external utilities. Finally, t e

Pollution Index Program is called to carry out pollution analysis to measure t e environmental impact of t e process and to direct c anges to reduce waste generation.

Subsequent c apters describe applying t is met odology in analyzing Motiva refinery alkylation process.

53

CHAPTER 4 MOTIVA ALKYLATION PROCESS Advanced Process Analysis System was applied to a commercial alkylation process at Motiva Refinery Complex in Convent, Louisiana. This chapter describes the

Motiva alkylation process and the simulation of the process developed with APAS.

Also, the validation of the process simulation is given, demonstrating that the simulation predicts the performance of the plant.

4.1 Process Description

Motiva Alkylation process is a 15,000 BPD STRATCO Effluent Refrigerated

Alkylation Plant. The heart of the process is the STRATCO reactor or contactor, which contacts the reactants in a high velocity propeller stream and removes heat from the exothermic reaction.

In the STRATCO Effluent Refrigerated Alkylation process, light olefins

(propylene, butylenes) are reacted with isobutane in the presence of sulfuric acid catalyst to form hydrocarbon, mainly in the iC7 to iC8 range, called alkylate. The alkylate product is a mixture of gasoline boiling range branched hydrocarbons which is blended with the refinery gasoline pool to increase the gasoline octane.

Process flow diagrams representing Motiva alkylation process are shown in

Figures 4.1, 4.2 and 4.3. These were prepared from P&ID’s of the plant. The reaction section of the process is shown in Figure 4.1. The refrigeration, depropanizer and deisobutanizer sections are shown in Figure 4.2. The saturate deisobutanizer section of the alkylation process is shown in Figure 4.3.

The process has four reactor pairs and four acid settlers as shown in Figure 4.1.

In the reaction section there are three feed streams, the olefin feed (HC01), the

54

isobutane feed (HC03) and the recycled olefin/isobutane mixture (HC32). The olefin feed contains the light olefins that are reacted with isobutane in the alkylation unit's

STRATCO stirred reactors. The isobutane stream is in excess to fully react with all of the olefins being charged to the unit. The process units are described in this section.

4.1.1 STRATCO Contactor

The two feed streams (HC01 and HC03) are cooled by exchanging heat with the net Contactor effluent stream (5E-628, 5E-628, 5E-630) from the suction trap/flash drum (5C-614) as shown in Figure 4.1. The two streams are then combined and fed to the STRATCO Contactors (5C-623 to 5C-630). The recycled olefin/isobutane stream is fed to the Contactors separately (HC34, HC38, HC41, HC45).

The STRATCO Contactor, in which the alkylation reaction occurs, is a horizontal pressure vessel containing a mixing impeller, an inner circulation tube and a tube bundle to remove the heat generated by the alkylation reaction as shown in Figure 4.4.

The feed is injected into the suction side of the impeller inside the circulation tube. The impeller rapidly disperses the hydrocarbon feed with acid catalyst to form an emulsion.

The emulsion is circulated by the impeller at high rates within the Contactor. The sulfuric acid present within the reaction zone serves as a catalyst to the alkylation reaction. The acid absorbs a small amount of hydrocarbon from side reactions and feed contaminants which reduces the sulfuric acid concentration.

55

1 M-3 5 M-4 Olefins Feed M-2 HC28 HC31 HC26 HC24 C301 HC27 HC25 C401 5C-614 Acid Settler Acid Settler HC01 5E-628 5C-631 5C-632 STFD HC30 AC05 AC15 C402 AC23 5E-629, HC32 AC12 HC02 Fresh Acid 630 M-11 M-24 M-1 AC02 M-7 3 HC03 HC04 5E-633 S-5 S-7 HC08 HC14 HC06 C403 AC07 AC18 S-2 HC07 AC09 HC11 AC20 3’ 5C-625 2 HC29 5C-623 Isobutane R1 R29 STRATCO 4 Reactor R20 R3 R2 R7 R6

S-19 HC34 HC38 HC33 S-23 M-15 HC23 R10 HC22

Acid Settler Acid Settler 5C-633 5C-634 HC40 AC26 AC37 Spent Acid AC34 AC45 M-13 M-17 AC23 S-11 HC19 AC40 AC29 AC42 5C-627 AC31 HC16 5C-629 HC14 R19

R12 R11 R16 R15

S-27 HC41 HC45

Figure 4.1. Reactor Section of the Alkylation Process

56

Alkaline 5E-634-639 , Compressor Water 5 5E-655, 656 5E-641-644 C301 C306 Fuel Gas C313A 1 C310 C312 5K-601 C307 C312A C303 C313 C315 Compressor KO Refrigerant Drum C310 Accumulator 5C-618 5C-658 5C-622 5C-615 Economizer Depropanizer Depropanizer Drum C302 C308 Charge Drum Charge 4 Caustic Settler 5C-616 C309 C312B C314 C314A 5E-640

C311 Discharge Water 5 C315 5E-610-653 5E-613 To FCCU, Fuel Gas C325 6 C316 C325 C318 5C-604 7 C329 5E-611 Depropanizer S-23 C329A C327 5E-621-624 C417 5C-603 IC4 from C319 storage 5E-657, 658 i-butane to C415 Storage C332 C317 Propane S-34 C414 5C-607 LPG C320 C328 Alkylate C321 Deisobutanizer C331 5E-612 3 C322 M-33 5E-627, 647 Fresh Caustic C406 C412 C415 S-41

C404 n-butane 2 5C-606 C410 C403 C407 C411 C404 C418 5E-616 M-34 C411 S-42 5C-660 5C-608 To IC4 Flush Pump C405A C427 C406 5E-696 C405 C409 C409 C410 C419 5E-626 C405 5E-695 Discharge Mixer C408 5E-617-620 C425 C407 C408 3’ Wash Water Wash Water Alkylate

Figure 4.2. Refrigeration, Depropanizer and Deisobutanizer Sections of the Alkylation process

57

SC406

6

5E-605 - 608

Saturate DIB Reflux Accumulator Saturate 5C-602 DeIsobutanizer Saturate Saturate Feed 5E-601 SC407 SC408 i-butane to storage SC401 SC402 5C-601 SC411 SC409 SC410 SC412

5E-609A SC404 SC413

5E-602 5E-603 7

n-butane to storage SC403

SC405

Figure 4.3. Saturate Deisobutanizer Section of the Alkylation Process

58

In order to maintain the desired spent acid strength (89.0 wt% H2SO4) in the last two of the eight Contactors, fresh acid (98.5 wt% H2SO4) is continuously charged to the first two Contactors, and an equivalent amount of 89.0 wt% spent acid is withdrawn from the acid settler.

Figure 4.4. STRATCO Effluent Refrigeration Reactor (Yongkeat, 1996)

The hydrocarbon and acid are separated in four settlers 5C-631 to 5C-634 and the acid-free hydrocarbon phase (HC22, HC23, HC25, HC27) flows from the top of the acid settler through a backpressure control valve. Part of it goes into the tube side of the contactor tube bundle (R1). The backpressure control valve is set at a pressure (60.0 psig) to maintain the contents of the settler in the liquid phase. As the hydrocarbon stream passes through the control valve, its pressure is reduced to about 5.0 psig, flashing a portion of the stream's lighter components, thereby, cooling the stream to a temperature of about 30 °F. When the two-phase stream passes through the tube bundle, additional vapor is generated as a result of absorbing heat generated by the alkylation reaction.

4.1.2 Refrigeration Section

As shown in Figure 4.1, the second portion (HC29) of the hydrocarbon effluent from the settler goes to the isobutane chiller (5E-633) and is used to reduce the temperature of the isobutane feed stream. It then combines with the Contactor effluent

59

stream leaving the tube bundle and flows into the suction trap side of the suction trap/flash drum (5C-614), where the liquid and vapor portions of this stream are separated. The suction trap/flash drum is a two-compartment vessel with a common vapor space. The net Contactor effluent is accumulated on one side of a separation baffle and is pumped to the effluent treating section. The cold refrigerant condensate is accumulated on the other side of the baffle. This effluent recycle stream consists mostly of isobutane and is returned to the Contactor by the effluent recycle pump. The vapor portions of both of these streams combine and flow to the suction of the refrigeration compressor (5K-601) shown in Figure 4.2.

The compressor driven by an electric motor, increases the pressure of the refrigerant vapor to allow condensing by cooling water. The vapor is condensed in two steps to provide a higher concentration of propane in the depropanizer feed.

Liquid generated by the first bank of condensers (e.g. 5E-634-639) is separated from the propane-rich vapor in the refrigerant accumulator (5C-615), further cooled by the refrigerant cooler, and then flashed across a control valve. Vapor and liquid are separated in an economizer (5C-616). The economizer vapor returns to an intermediate stage of the compressor (5K-601) to be recompressed, while the liquid is returned to the suction trap/flash drum (5C-614).

The remaining vapor portion of the compressor discharge stream (C310) flows from the separator to the "total condenser" (5E-641-644). Liquid from the total condenser is accumulated in the Depropanizer Charge Drum (5C-618), and ultimately fed to the Depropanizer (5C-603) for stripping of propane before being sent to the

Alkylate Deisobutanizer (5C-606).

60

4.1.3 Depropanizer (5C-603)

The depropanizer feed stream is pumped from the charge drum to the depropanizer feed caustic wash (5C-658) to neutralize acidic components. The neutralized stream then flows to the depropanizer feed/bottoms exchanger (5E-610-5E-

653) before entering the tower, as shown in Figure 4.2.

The overhead propane vapor from the depropanizer (C325) is totally condensed by cooling water and collected in the depropanizer reflux accumulator drum (5C-604).

A part of the liquid is refluxed to the column's top tray (C329), and the rest is further cooled and sent to the LPG section (C328). The depropanizer bottoms product (C317) temperature is reduced in the feed/bottoms exchanger (5E610-653). The stream then enters a water-cooled exchanger (5E-611) before it is split into three portions, one each for the Alkylate Deisobutanizer, the Contactor ( 3 ) and storage.

4.1.4 Alkylate Deisobutanizer (5C-606)

The net effluent from the suction trap/flash drum (5C-614) is heated in the feed/effluent exchangers (5E-628; 5E-629, 5E-630) as shown in Figure 4.1 and then treated with caustic to remove any traces of acid (5C-660 and 5C-608 in Figure 4.2).

The treated net effluent is fed to the alkylate deisobutanizer (DIB) tower (5C-606) to strip isobutane from alkylate as shown in Figure 4.2. Two steam-fired thermosiphon reboilers (5E-695, 5E-696) supply the majority of the heat to the lower portion of the column by vaporizing liquid side draws (C411) from the column as well as the DIB bottoms liquid (C408). A portion of the depropanizer bottoms is fed to the column as the reflux (C322).

Normal butane is removed as a vapor side-draw (C412). The vapor is condensed and cooled by cooling water (5E-627, 5E-647). The normal butane product (C415) can

61

be used for vapor pressure blending or sold. The rate of the normal butane product will vary depending on the desired deisobutanizer alkylate product Reid Vapor Pressure

(RVP).

The overhead isobutane vapor is condensed by cooling water and accumulated in the isobutane accumulator pot (5C-607). Make-up isobutane (C417) from storage is added to the pot and the combined stream (C418) is sent to the Contactors to react with the olefins ( 3’ ).

The deisobutanizer bottoms (C405) is the alkylate product. It is cooled by the deisobutanizer feed/effluent exchanger (5E-616) and also a heat exchanger with cooling water (5E-617-5E-620). The cooled alkylate (C407) is removed as a product and sent to storage.

4.1.5 Saturate Deisobutanizer (5C-601)

Another source for isobutane is the saturate deisobutanizer (SatDIB) column (Figure

4.3), which strips isobutane from saturate feed coming from reformer unit of the refinery (SC401). The saturate feed, containing mainly C4’s, flows to the SatDIB feed/bottoms exchanger (5E-601) before entering the tower, as shown in Figure 4.3.

The overhead isobutane vapor is condensed by cooling water and a fraction of it (stream

SC413) is accumulated in the isobutane accumulator pot (5C-607, Figure 4.2). The saturate deisobutanizer bottoms (SC403) are mainly n-butane product. It is cooled by the SatDIB feed/bottoms exchanger (5E-601) and also by a heat exchanger with cooling water (5E-603). The cooled n-butane (SC405) is sent to storage.

4.2 Process Simulation

This section describes the models of the various process equipments in the alkylation process. The process flow diagram, as developed with Flowsim, the

62

flowsheet simulation tool of Advanced Process Analysis System, is shown in Figure

4.5.

The equipment in the plant can be categorized according to their functions as follows:

Reaction Zone:

• STRATCO Contactor

• Acid settler

Separation Zone:

• Depropanizer

• Alkylate Deisobutanizer

• Saturate Deisobutanizer

• Suction Trap Flash Drum

• Economizer

• Compressor

Heat Transfer Zone:

• Heat Exchangers

• Condensers

• Reboilers

Miscellaneous:

• Mixers

• Splitters

• Reflux Accumulators

The models are developed for each of these units and are given in this chapter.

63

The alkylation process involves mixtures of a variety of components. The components and compositions of streams vary widely from one section to the other. For record keeping, all the hydrocarbon streams in the plant are composed of the following components:

1. C3- Propane and lower

2. C4= Butenes

3. iC4 iso-butane

4. nC4 normal-butane

5. iC5 iso-pentane

6. nC5 normal-pentane

7. iC6 iso-hexane

8. iC7 iso-heptane

9. iC8 iso-octane

10. iC9+ iso-nonane and higher

All the sulfuric acid streams in the plant are composed of:

11. H2SO4 sulfuric acid

12. Water and impurities

The streams carrying the reaction products from the contactors to the acid settlers contain all of the twelve components.

64

Figure 4.5. Process flow diagram, as developed with the Flowsheet Simulation tool of Advanced Process Analysis System

65 4.2.1 STRATCO Contactor (5C-623)

The sulfuric acid alkylation reaction mechanism for pure propylene and pure butylene feeds were shown in Tables 2.2 and 2.3. These are based on the Schmering carbonium ion mechanism with modification introduced to account for iC9 and iC10 formation. The material balance on reactants and the product formation equations are shown in the same tables. The parameters that affect the performance of the contactor are: AC07 1. Operating Temperature

2. Acid Strength HC07

3. Isobutane Concentration AC09

R2 4. Olefin Space Velocity R3

HC34 Figure 4.6: Contactor 5C-623

The contactor was modeled as of a Continuously Stirred Tank Reactor (CSTR) with heat exchange. The intense mixing provided by the propellers ensures that there is no spatial variation in concentration, temperature, or reaction rate throughout the vessel.

The composition and temperature of the exit stream were the same as those inside the reactor.

Referring to Figures 4.1 and 4.6, the material and energy balance equations for the contactor 5C-623 are as follows:

Description

Inlet streams: HC07 (isobutane- olefin mixture), HC34 (effluent recycle-mainly

isobutane), and AC07 (sulfuric acid catalyst)

Outlet streams: AC09

Coolant streams: R2 (in), R3 (out)

Material Balance

Overall:

Mass flowrate in – Mass flowrate out = 0 (4.1)

FHC 07 + F HC 34 + F AC 07 − F AC 09 = 0 (4.2)

FR2 − FR3 = 0 (4.3)

where, FX is the total mass flow rate (metric ton/min) of stream X.

Component:

Mass flowrate in – Mass flowrate out + Mass generation rate = 0 (4.4)

i i i i a i FHC 07 + FHC 34 − FAC 09 + r V 5C 623 MW = 0 (4.5) for the hydrocarbon components. Here, r i is the reaction rate of component i, in terms of moles of i produced per unit time, per unit volume of the acid catalyst (metric ton moles/(m3⋅min)), and MW i is the molecular weight of component i. The quantity

a V5 C 623 , shown in the material balance equations, is the volume of acid in the contactor, and is typically 60 % of the total contactor volume.

Considering the degradation of the catalyst, the material balance equation for the acid component can be written as

i i a FAC 07 − FAC 09 − R 5C 623 = 0 (4.6)

a where, R5 C 623 is the rate of degradation of the acid (metric ton/min). The acid degradation rate can be related to the volumetric flowrate of butylenes into the contactor as follows (Graves 1999),

67

a C 4= R5 C 623 = 0.121Q HC 07 (4.7)

The total mass flow rates are calculated as the sum of the component flow rates.

i ∑ FX − FX = 0 (4.8)

The subscripts to the process variables like mass flow rates, specific enthalpies and stream temperatures indicate the stream names, and their superscripts indicate the component. The subscripts to the process parameters indicate the process unit they represent.

The material in the contactor is a two-phase mixture, consisting of a hydrocarbon phase and an acid phase. Since the reaction model assumes that all the reactions occur in the acid phase, the concentrations shown in the reaction model equations are those in the acid phase.

The olefin feed to the process is composed of about 45% lv butylenes (1-BUT +

C2B + T2B) and about 3.5% lv propylene among other paraffins. Because of the low presence of propylene and the non-participation of paraffins in the alkylation reaction, the reaction mechanism of alkylation of isobutane with pure butylene (Table 2.3) was adopted. The reaction rate terms in Equation 4.5 is evaluated using corresponding equations from Table 2.3. Appendix A shows the procedure followed for calculating the concentrations of the contactor components in the acid phase.

Alkylation of isobutane with olefins is an exothermic reaction. The reaction mixture must be maintained in a particular temperature range (0-10 °C) in order to obtain the desired yields. Higher operating temperatures result in oxidation and polymerization reactions, decreasing the yields. Lower temperatures result in decreased

68

effectiveness due to the increase in acid viscosity and decrease in solubility of hydrocarbons in the acid phase. Therefore, the temperature in the contactor is maintained by recycling the cold refrigerant condensate from the Suction Trap Flash

Drum, which consists mostly of isobutane, and by circulating a part of the flashed, acid- free effluent from the settlers, through the contactor tube bundle. When the two-phase stream passes through the tube bundle, additional vapor is generated.

The tube bundle in the contactor, is effectively, a partial boiler, which vaporizes a portion of the coolant. This can be modeled by assuming that the two-phases in the coolant mixture are in equilibrium with each other throughout the tube bundle. The component mass balance for the two-phase coolant stream can be written as

i i i i (FR 2l+ F R2v) − (F R3l+ F R3v ) = 0 (4.9)

yi Ki = where, ∑ xi = 1, ∑ yi = 1 (4.10) xi

In the above equation, yi is the mole fraction of component i in the vapor phase, and

xi is that in the liquid phase.

The model can be simplified by assuming that all the vapor generation in the tube bundle, due to transfer of heat from the contactor, occurs at a point outside the tube bundle; i.e. after the coolant has passed through it. This is the same as considering the partial boiler to be equivalent to a heater followed by an isothermal flash operation.

Then, the mole fractions can be written as

i i FR3v FR3l MW i MW i y i = i and xi = i (4.11) FR3v FR3l ∑ MW i ∑ MW i

69

Energy Balance

Overall:

Energy in + Energy generated - Energy out = 0 (4.12)

hHC 07 + hHC34 + hAC 07 + hR2 + H 5C 623 − hAC 09 − hR3 = 0 (4.13)

where, h X is the enthalpy of stream X in (MJ/min), and H 5C 623 is the heat generated in the contactor by the exothermic reaction (MJ/min).

The stream enthalpies are calculated from the component specific enthalpies, as

i i hX = ∑ FX hX for single phase streams, and (4.14)

i i i i hX = (∑ FX hX )l + (∑ FX hX )v (4.15)

for two-phase streams (R2 and R3)

i where, hX is the specific enthalpy of the component i in stream X in (MJ/metric ton).

Heat Transfer:

Energy in – Energy out – Energy transferred out = 0 (4.16)

hR2 − hR3 −U 5C 623 A5C 623ΔTlm = 0 (4.17)

2 where, U 5C 623 is the overall heat transfer coefficient (MJ/m ⋅°C⋅min), A5C 623 is the

2 area of heat transfer (m ), and ΔTlm is the log mean temperature difference in the contactor 5C-623 (°C).

(T − T )− (T − T ) ΔT = 5C 623 R2 5C 623 R3 (4.18) lm (T − T ) ln 5C 623 R2 (T5C 623 − TR3 )

70

The mass and energy balance equations for contactor 5C-623 are summarized in Table

4.1.

Table 4.1. Summary of the contractor model

Material Balances

Overall FHC 07 + FHC 34 + FAC 07 − FAC 09 = 0

FR 2 − FR3 = 0

i i i i a i Species FHC 07 + FHC34 − FAC 09 + r V5 C 623 MW = 0

i' i' a FAC 07 − FAC 09 − R5C 623 = 0

a C 4= R5C623 = 0.121QHC07

where, i= 1,2,3,4,5,6,7,8,9,10 and i’= 11

Energy Balances

h HC07 + h HC34 + h AC 07 + h R2+ H 5C 623 − h AC 09 − h R3 = 0

h R2 − h R3−U 5C 623 A 5C623 ΔTlm = 0

The equations given in Table 4.1 describe each of the four pairs of contactors.

However, the kinetic model is not a function of the catalyst concentration, and plant data are not available for sulfuric acid concentration in the contactors. Consequently, the model for each of the contactors is the same.

71

4.2.2 Acid Settler (5C-631)

The acid settler separates the emulsion from the contactor into the acid and hydrocarbon phases. One acid settler is provided to each pair of contactors. The inputs to the settler are the effluents from the contactor pair and the outputs are the separated acid and hydrocarbon streams.

The material and energy balance equations for the acid settler (5C-631) are as shown in Table 4.2. The mass flow rate of each component coming in is equal to the mass flow rate of each component leaving of the settler. Since there is no heat transferred from or into the settler, the temperatures of the various input and output streams can be assumed to be equal. We also assume that the separation of the two phases in the settler is complete, so that there is no acid component in the output hydrocarbon stream and vice-versa.

Table 4.2. Summary of the acid settler model

Material Balances

i i FAC 09 − FHC 27 = 0 i=1,2,3,4,5,7

11 12 F AC 09 + F AC 09 − F AC 05 − F AC12 = 0

11 11 F AC 05 − sC 631 F AC 09 = 0

12 12 F AC 05 − sC 631 F AC 09 = 0

11 11 x AC 05 − x AC12 = 0

Energy Balances

T AC 09 = T HC 27 ; T AC 09 = T AC12 ; T AC 09 = TAC 05

72

4.2.3 Depropanizer (5C-603)

The Depropanizer was modeled using Smith-Brinkley Group method (Smith,

1963; Perry et al., 1984). This method can be applied to absorption, extraction processes and distillation. The equation, which applies to the distillation process, is for each component i,

N −M (1+ S n,i )+ R(1− S n ) f = N −M N −M M +1 (4.19) (1+ S n,i )+ R(1− S n) + hS n (1− S m )

i i where R is the external-reflux ratio FC329/FC328 and f = (F C317/F C316), is the fraction of i leaving in the bottoms product. The quantity S is the stripping factor (for each component) and is defined for each group of stages in the column by Sn,i=KiV/L and

Sm,i=K'iV'/L'. K is the equilibrium ratio y/x for each component and V, L are the vapor and liquid molar flow rates in the column. The Ki, V and L values are the effective values for the top column section; K'i, V' and L' are the effective values for the section below the feed stage. The quantity h depends on whether K or K' is used for the feed stage. If the feed is mostly liquid, the feed stage is grouped with the lower stages and

K'i L ⎛ 1− Sn ⎞ h = ⎜ ⎟ (4.20) i K L' ⎜1− S ⎟ i ⎝ m ⎠i

If the feed is mostly vapor,

L ⎛ 1− S ⎞ h = ⎜ n ⎟ (4.21) i L' ⎜1− S ⎟ ⎝ m ⎠i

The following four specifications for the column are necessary:

1. N, the total number of equilibrium stages

2. M, the number of stages below the feed stage

3. The reflux rate or the maximum vapor rate at some point in the column

73

4. The distillate rate D

Specification of D and L (or V), along with specifications of the feed stream variables fixes the assumed phase rates in both sections of the column. Determination of separation factors Sn and Sm then depends upon estimation of individual K values. If ideal solutions are assumed, K values are functions of only temperature and column specified pressure. Estimation of K values and, in turn, Sn and Sm values for each component reduces to estimation of the effective temperature in each column section or group of stages.

The effective temperature is assumed to be the arithmetic average for simplicity.

(T + T ) T = N M +1 (4.22) n 2

T + T T = ( M +1 1 ) (4.23) m 2

The feed enters the column at the (M+1)th stage. The material and energy balance equations for the depropanizer are shown in Table 4.3. The Saturate Deisobutanizer column is modeled in a similar way.

74

Table 4.3. Summary of the depropanizer model

Material Balances

Overall FC 316 + FC 329 − FC 317 − FC 325 = 0

i i i i Species FC 316 + FC 329 − FC 317 − FC 325 = 0

where i=1,3,4,5

Smith- Brinkley Method Equations

K i= f (P C 603 ,Tn )

' K i= f (P C 603 ,Tm )

KFi C325 S n,i = FC329

K 'V ' S = i m,i L'

N −M (1+ S n,i )+ R(1− S n,i ) f i = N −M N −M M +1 (1+ S n,i )+ R(1− S n,i )+ h i S n,i (1− S m,i )

i FC 317 f i = i FC316

K' L ⎛ 1− S ⎞ h = i ⎜ n ⎟ i KL'⎜ 1− S ⎟ i ⎝ m ⎠i

where i=1,3,4,5,7

i i xC 317 − xC 323 = 0 , where i=1,3,4,5

TC 317 − TC 323 = 0

75

4.2.4 Alkylate Deisobutanizer (5C-606)

The alkylate deisobutanizer column has two feed streams, two product streams, two side streams and two reboilers, one of which is used as a side reboiler. One of the feed streams is fed to the column, at the top tray, in place of the reflux. There was no readily available short cut method to model such a distillation column as it is, therefore, the column was divided into three reasonable sections (C-606A, C-606C and C-606D) which can be modeled using Smith Brinkley Method and by employing equilibrium stage assumption. General Smith-Brinkley Method is used for a classical distillation column with recycle and reboiler streams, however the model of deisobutanizer requires the derivation of the method for the rectifying and stripping sections separately.

Derivations can be found in (Smith, 1963) – Design of Equilibrium Stage Processes.

The equations for the alkylation deisobutanizer are given in Table 4.4.

76

Table 4.4. Summary of the deisobutanizer model

Material Balances

Overall FC 404 + FC 432 + FC 322 − FC 414 − FC 430 = 0

FC 430 + FC 427 − FC 431 − FC 425 = 0

FC 426 − FC 428 − FC 405 = 0

FC 427 − FC 431 = 0

FC 425 − FC 430 = 0

i i i i i Species FC 404 + FC 432 + FC 322 − FC 414 − FC 430 = 0

i i i i FC 430 + FC 427 − FC 431 − FC 425 = 0

i i i FC 426 − FC 428 − FC 405 = 0

where i=1,3,4,5

Smith- Brinkley Method Equations

K i,C 606 A = f (PC 606 A ,Tn,C 606 A )

' Ki,C606 A = f (PC 606 A ,Tm,C 606 A )

K i,C 606C = f (PC 606C ,TC 425 )

' Ki,C 606D = f (PC 606D ,Tm,C606D )

' ' K i,C 606 AV C 606 A S m,i,C 606 A = ' LC 606 A

KFi,C 606 A C 414 S n,i,C 606 A = FC 322

KFi,C 428 C 428 S m,i,C 606 D = FC 426

77

Table 4.4. Summary of the deisobutanizer model (Cont’d)

Smith- Brinkley Method Equations

N −M S N −M (1− S n,i,C 606 A)+ q i,C 606 A(S n,i,C 606 A− S n,i,C 606 A ) + qF h S N −M (1− S M ) i,C 606 A i,C 606 A n,i,C 606 A m,i,C 606 A f i,C 606 A = N −M N −M M +1 (1− S n,i,C 606 A)+ h i,C 606 A S n,i,C 606 A(1− S m,i,C 606 A )

i FC 430 f i,C 606 A = i i i (FC 404 + FC 322 + FC 432 )

F ⎛ 1− S ⎞ h = C322 ⎜ n,C606 A ⎟ i,C 606 A L' ⎜1− S ⎟ C 606 A ⎝ m,C 606 A ⎠i

F K ' (S M D−1 −1)+ F (S −1) M C 428 i,C 606D m,i,C 606D C 426 m,i,C 606D x i,C 606D = 2 FC 426 (S m,i,C 606D −1)

where i=1,3,4,5,7

i i i xx C 425 K C 606C− xx C 431 = 0 , where i=1,3,4,5

i i xC 408 − xC 405 = 0 , where i=1,3,4,5

' LC 606 A= F C 322 + qC 606 A F C 404

' V C 606 A= F C 432

TC 431 − T C 425 = 0

78

4.2.5 Suction Trap/Flash Drum (5C-614)

The suction trap/flash drum (STFD) splits the hydrocarbon effluent from the settlers based on volatility. It is a two-compartment vessel, with a common vapor space.

The net contactor effluent is accumulated on one side of the baffle and the cold refrigerant condensate on the other. The vapors from the two compartments combine, and flow out from the top.

The STFD is modeled as two C301 C302 separate units – suction trap and flash drum, x with the vapor streams from the two units HC31 C311 C401 HC32 combining in a mixer, as shown in Figure Suction Trap Flash Drum 4.7. The suction trap acts as the separating Figure 4.7: STFD 5C-614 vessel for the two-phase contactor from the contactor tube bundles and from the isobutane chiller (5E-633). The liquid and vapor streams leave the vessel as two separate streams. In the Flash drum, the cold condensate from the economizer is flashed and separated. This unit is modeled as an equilibrium stage with adiabatic flash. The material and energy balance equations for the two units are as follows:

Suction Trap:

Overall material balance: (FHC 31l + F HC 31v) − (F C 401 + FC 301 ) = 0 (4.24)

FHC 31l− F C 401 = 0 (4.25)

i i i i Component material balance: (FHC31l + F HC31v)− (F C 401 + FC 301 )= 0 (4.26)

i i FHC31l− F C 401 = 0 (4.27)

79

Energy balance: THC 31 = TC 301 = TC 401 (4.28)

Flash Drum

Overall material balance: FC 311 − FHC 32 − FC 302 = 0 (4.29)

i i i Component material balance: FC 311 − FHC 32 − FC 302 = 0 (4.30)

yi Ki = , where, ∑ xi = 1, ∑ yi = 1 (4.31) xi

Energy balance: hC 311 − h HC 32 − hC 302 = 0 (4.32)

The flash operation has been described, in detail, in the contactor modeling section. The

Suction Trap Flash Drum model described above is summarized in Table 4.5.

80

Table 4.5. Summary of the suction trap/flash drum model

Material Balances

Overall (FHC 31l + F HC 31v) − (F C 401 + FC 301 ) = 0

FHC 31l− F C 401 = 0

FC 311 − FHC 32 − FC 302 = 0

i i i i Species (FHC 31l + F HC31v)− (F C 401 + FC 301 )= 0

i i FHC31l− F C 401 = 0

i i i FC 311 − FHC 32 − FC 302 = 0 , where i=1,3,4,5,7

yi Ki = , where, ∑ xi = 1, ∑ yi = 1 xi

Energy Balances

THC 31 = TC 301 = TC 401

hC 311 − h HC 32 − hC 302 = 0

81

4.2.6 Economizer (5C-616)

The Economizer separates the vapor and liquid C310 phases generated when the liquid portion of the C309 propane-rich vapor, in the Depropanizer section, is flashed. This unit is modeled the same way as the flash C311 drum of STFD, i.e. assuming an adiabatic flash Figure 4.8. Economizer 5C-616 operation, where the two phases generated are in equilibrium with each other. The mass and enthalpy balances are:

Overall material balance:

FC309 − FC 310 − FC 311 = 0 (4.33)

Component material balance:

i i i FC 309 − FC 310 − FC 311 = 0 (4.34)

yi Ki = , where, ∑ xi = 1, ∑ yi = 1 (4.35) xi

In the above equation, K i is the distribution coefficient or the K-factor, yi is the mole

fraction of component i in the vapor phase, and xi is that in the liquid phase.

Enthalpy balance:

hC 309 − hC 310 − hC 311 = 0 (4.36)

where, hX is the enthalpy of stream X.

4.2.7 Compressor (5K-601)

The combined vapor stream from Suction Trap and the Flash Drum is rich in propane, and is sent to the depropanizer section. The vapor is compressed in the

82

Refrigeration compressor (5K-601), to allow condensing by cooling water. The compressor was assumed to operate under adiabatic conditions, and hence the calculations were based on the following formulas (Perry et al., 1984):

Adiabatic head is expressed as

⎡(k −1) ⎤ k ⎛ p ⎞ k H = RT ⎢⎜ 2 ⎟ − 1⎥ (4.37) ad 1 ⎢⎜ ⎟ ⎥ k − 1 ⎝ p1 ⎠ ⎣⎢ ⎦⎥

where H ad = adiabatic head, ft; R = gas constant, (ft⋅lbf)/(lb⋅°R) = 1545/molecular

2 weight; T1 = inlet gas temperature, °R; p1 = absolute inlet pressure, lbf/in ;

2 p2 = Absolute discharge pressure, lbf/in , k = ratio of specific heat capacities ( ccp v ).

The work expended on the gas during compression is equal to the product of the adiabatic head and the weight of the gas handled. Therefore, the adiabatic power is as follows:

⎡ (k −1) ⎤ WH k WRT ⎛ p ⎞ k hp = ad = 1 ⎢⎜ 2 ⎟ − 1⎥ (4.38) ad ⎢⎜ ⎟ ⎥ 550 k − 1 550 ⎝ p1 ⎠ ⎣⎢ ⎦⎥

⎡ (k −1) ⎤ k ⎛ p ⎞ k or hp = 4.36 ×10 −3 Q p ⎢⎜ 2 ⎟ − 1⎥ (4.39) ad 1 1 ⎢⎜ ⎟ ⎥ k − 1 ⎝ p1 ⎠ ⎣⎢ ⎦⎥

where hpad = power, hp; W = mass flow, lb/s; and Q1 = volume rate of gas flow, ft3/min.

Adiabatic discharge temperature is:

83

(k −1) ⎛ p ⎞ k ⎜ 2 ⎟ (4.40) T2 = T 1 ⎜ ⎟ ⎝ p1 ⎠

C303 C304

C310

Figure 4.9: Compressor 5K-601

The constraint equations, for the compressor, are developed using the above formulas.

Overall material balance:

FC 303 + FC 310 − FC 304 = 0 (4.41)

Component material balance:

i i i FC 303 + FC 310 − FC 304 = 0 (4.42)

Enthalpy balance:

hC 303 + hC 310 + hp ad − h C 304 = 0 (4.43)

(k −1) ⎛ p ⎞ k ⎜ 2 ⎟ (4.44) T2 = T1 ⎜ ⎟ ⎝ p1 ⎠

where hpad is the compressor as given in Equations (4.38, 4.39). The constraints of the compressor are summarized in Table 4.6.

Table 4.6. Summary of the compressor model

84

Material Balances

Overall FC 303 + FC 310 − FC 304 = 0

i i i Species FC 303 + FC 310 − FC 304 = 0 where i=1,3,4,5

Energy Balances

hC 303 + hC 310 + hp ad − h C 304 = 0

(k −1) ⎛ p ⎞ k ⎜ 2 ⎟ T2 = T1 ⎜ ⎟ ⎝ p1 ⎠

⎡ (k −1) ⎤ k 8.314F T ⎛ p ⎞ k hp = C 306 C303 ⎢⎜ 2 ⎟ −1⎥ ad ⎢⎜ ⎟ ⎥ k −1 55 ⎝ p1 ⎠ ⎣⎢ ⎦⎥

4.2.8 Olefin Feed Effluent Exchanger (5E-628)

The olefin feed effluent exchanger cools the olefin feed by exchanging heat with the alkylate effluent stream from the suction trap/flash drum. The material and energy balance equations for this exchanger are shown in Table 4.7. Since no mass transfer is involved in this unit, the masses of the two streams are conserved. The heat transferred between the streams can be accounted for as shown in the energy balance equations.

Table 4.7. Summary of the exchanger model

85

Material Balances

Overall FHC 01 + FHC 02 = 0

FC 401 + FC 402 = 0

i i Species xHC01 + xHC02 = 0

i i xC 401 + xC 402 = 0

where i= 1,2,3,4,5

Energy Balances

(h HC 02 − hHC 01 ) − (hC 401 − hC 402 ) = 0

(h HC 01 − h HC 02 ) −U 5E 628 A 5E628 ΔTlm = 0

(T − T )− (T − T ) ΔT = HC 02 C 401 HC 01 C 402 lm (T − T ) ln HC 02 C 401 (THC 01 − TC 402 )

86

4.3 Model Validation

The accuracy of the alkylation process model to predict the performance of the plant has to be established. This model validation was performed by solving the data validation problem using a set of plant data (steady state operation point #1) and then comparing the results of the model with the plant data. For the 125 measured plant variables, 88 were within the accuracy of the measurements. The remaining 37 variables are shown in Table 4.8 for the plant data from the DCS, their reconciled value and the standard measurement error ∈i (∈i = |yi -xi |/σi, where is σi the standard deviation of the process variable). In consulting with the alkylation plant process engineers, it was concluded that these 37 variables were within the range of possible process values. This includes FC316, FC322, x5C417, xx5C412, xx7C414. The discrepancy in the variable xx1C322 is a result of the model assumption that the propane purge from the drum 5C-

618 occurs infrequently. In summary, the model of the process accurately predicts its performance and therefore can be used for on-line optimization.

Table 4.8. Plant vs. Model Data

Variable Name Plant Data Reconciled Data from Standard (yi) Data Validation Measurement (xi) Error (∈i) FAC02 0.1125 0.1600 4.2235 FAC12 0.1259 0.1600 2.7085 FAC23 0.1253 0.1600 2.7653 FAC45 0.1040 0.1600 5.3846 FC308 2.1990 3.1032 4.1120 FC316 0.6581 1.8000 17.3515 FC322 0.4427 1.5619 25.2812 FC328 0.0942 0.0535 2.6399 FC403 3.8766 2.2834 4.1097 FC412 0.0324 0.0418 2.8968 FSC411 2.7287 1.3525 5.0436 FstmE612 0.1425 0.0889 3.7607

87

x1C417 0.0372 0.0255 3.1309 x2SC402 0.0136 0.0084 3.7929 x2SC408 0.0221 0.0002 9.9048 x3C325 0.0017 0.0000 10.0000 x3SC403 0.0103 0.0212 10.5665 x4C316 0.0580 0.0796 3.7155 x4SC408 0.0331 0.0088 7.3475 x5C316 0.0020 0.0060 19.8000 x5C417 0.0009 0.0295 286.2300 x5HC32 0.0096 0.0306 22.0134 x6SC402 0.0167 0.0666 29.8204 x6SC403 0.0250 0.0950 27.9946 x7HC32 0.0197 0.0497 15.2312 x7SC402 0.0022 0.0032 4.3956 x7SC408 0.0022 0.0000 10.0000 xx1C322 0.0027 0.1167 428.5338 xx1C414 0.0330 0.0800 14.2498 xx2HC01 0.4525 0.1291 7.1481 xx3C407 0.0003 0.0000 7.4194 xx3HC01 0.3558 0.0125 9.6498 xx4C407 0.1124 0.0853 2.4068 xx5C407 0.0803 0.1506 8.7555 xx5C412 0.0022 0.0581 255.6751 xx5C414 0.0021 0.0011 4.8325 xx7C414 0.0015 0.0080 44.4218

4.4 Summary

In this chapter, Motiva alkylation process and the system model for the process was described. Also the validity of the model was demonstrated by comparing plant data with model predictions. The complete model equations can be found in Appendix

B. The next chapter describes the results from the Advanced Process Analysis System.

88 CHAPTE 5 ESULTS Thi chapter de cribe the optimization of Motiva Alkylation proce u ing Advanced Proce

Analy i Sy tem. The methodology of the y tem, a de cribed in Chapter 4, i to develop proce

model u ing Flow heet Simulation program, obtain optimal etpoint for the di tributed control

y tem u ing On-Line Optimization program, and then to perform off-line analy i u ing Pinch

Analy i and Pollution A e ment program . The Reactor Analy i program wa u ed to develop the

kinetic model for the alkylation reaction .

5.1 Flowsheet Simulation

Developing a proce model u ing Flow heet Simulation program involve drawing the proce flow diagram, de cribing the proce u ing con traint equation , and pecifying mea ured and

unmea ured variable , con tant and parameter . Each of the e tep i explained with numerou

illu tration in the u er manual given in Appendix J.

5.1.1 Process Flow Diagram

Before u ing the program to enter proce detail , a model name and de cription have to be

pecified. The name ‘Alkyl’ wa u ed for the alkylation proce . All the information entered for the proce will be a ociated with thi name. The next tep in u ing the program i drawing the proce

flow diagram, which i a vi ual repre entation of the proce . The program ha an interactive graphical

interface to facilitate ea y drawing and editing of the flow diagram. It provide a toolbar with button

to draw unit and tream , move and re ize them, and change their appearance. Each unit and tream

in the flow diagram ha a unique name and de cription.

The proce flow diagram for Motiva alkylation proce , a drawn in the Flow heet

Simulation program, i hown in Figure 4.5. Thi i the program equivalent of the flow diagram

hown in Figure 4.1 to 4.3. For each proce unit, the material and energy balance equation are

89 added a de cribed in the u er’ manual in Appendix J. Al o, rate equation and equilibrium relation

are added a are thermodynamic function and global data. The e make up the con traint equation for

on-line optimization. In addition, an economic model of the proce i included (See Section 5.2.3).

5.1.2 Constraint equations

A de cribed in Chapter 3, con traint fall into two type : equality con traint and inequality

con traint . Equality con traint depict con ervation law (material and energy balance ), equilibrium

relation and empirical formula . Inequality con traint include minimum approach temperature ,

demand for product, availability of raw material and capacitie of proce unit . Equality con traint

for the alkylation proce were developed in Chapter 4. For example, Table 4.1 how the equality

con traint for contactor 5C-623. Con traint can be entered into the program at pecific unit or tream

level, or at global level. Con traint that pertain to a pecific unit or tream, ay, i obutane ma balance or temperature limit for contactor 5C-623, are entered by clicking on that unit or tream.

Con traint that pertain to multiple unit or tream , ay, the overall ma balance for the proce , are

entered at the global level. An optional caling factor can be pecified for equality and inequality

con traint . Refer to the program’ u er manual for a detailed explanation.

The Alkylation Proce Model, “Alkyl”, ha 1,579 equality con traint and 50 inequality

con traint (Appendix B) to de cribe 76 proce unit and 110 tream . The e are tored along with

other information of the proce , in the central databa e, which can be acce ed by all the program of

Advanced Proce Analy i Sy tem. The next tep in developing the proce model u ing Flow heet

Simulation program i to enter pecification : mea ured and unmea ured variable , plant parameter ,

con tant and enthalpy information. The following ection de cribe the e pecification for the

alkylation proce .

90 5.1.3 Measured Variables

A de cribed in Chapter 3, mea ured variable are proce variable who e value that can be

obtained from the plant’ di tributed control y tem. For the alkylation proce , information from the

di tributed control y tem wa reque ted by pecifying Tag Id of mea uring device elected from proce and in trumentation diagram . The data corre ponding to the Tag Id were made available in

the form of Micro oft Excel file , which contained plant mea urement obtained at interval of one

minute over a period of two con ecutive day for five individual date (11/08-09/98, 11/22-23/98,

12/06-07/98, 12/13-14/98 and 12/27-28/98). The e file contained variable uch a flow rate ,

temperature and pre ure . Al o, concentration from laboratory analy i were made available for

tho e date .

A mentioned in Chapter 3, the On-Line Optimization program input plant data at teady

tate for calculating optimal etpoint . So, the data from di tributed control y tem wa analyzed for

teady tate vi ually by u ing MathCAD’ graphic capabilitie . Thi information i given in

Appendix I. Six (6) range are identified where for all of the mea ured variable the fluctuation are

minimal. They are given in Appendix I. The average value for the e teady tate were entered a plant data in the Flow heet Simulation program (Appendix C.3). The proce model Alkyl ha 125

mea ured variable in the fir t two, 120 in the third, forth and fifth, and 113 in the la t teady tate data

et .

For each mea ured variable, the program tore all the information required to run all the program of Advanced Proce Analy i Sy tem. The information that i required for GAMS (General

Algebraic Modeling Sy tem) to olve the optimization problem i name of the variable, plant data and

tandard deviation. Other information like initial point, caling factor, lower and upper bound value

91 are provided by the u er. The e value entered into Flow heet Simulation program are given in

Appendix C.3.

5.1.4 Unmeasured Variables

A the name ugge t , unmea ured variable are tho e proce variable who e value cannot be obtained from the di tributed control y tem. Unmea ured variable pertaining to a pecific unit or

tream can be pecified in the Flow heet Simulation program opening the window for the unit or

tream. Tho e that pertain to multiple unit or tream , or to the entire proce , can be pecified a

global variable . The required information for pecifying an unmea ured variable i it name. Optional

information, like initial value, bound and caling factor, are defaulted by the program. The program’

u er manual provide a detailed de cription for thi tep. The alkylation proce model ha 1,509

unmea ured variable . Initial value for the e unmea ured variable entered into Flow heet Simulation program are given in Appendix C.4.

5.1.5 Parameters

A mentioned in Chapter 3, parameter are tho e quantitie in a proce model, which change

lowly with time, and are related to the degradation of performance of equipment. Parameter include coefficient of heat tran fer and fouling factor in heat exchanger , and cataly t deactivation parameter . The information required for pecifying parameter in the Flow heet Simulation program include it name, de cription, initial value, lower and upper bound , and unit . The program’ u er manual de cribe thi tep in detail.

The alkylation proce ha overall heat tran fer coefficient and LMTD correction factor of heat exchanger , reflux ratio of two di tillation column (5C-601 and 5C-603) and pre ure drop in two heat exchanger (5E-634 and 5E-640) a parameter . The alkylation proce model ha 64

92 parameter and the plant parameter value entered into Flow heet Simulation program are given in

Appendix C.5.

5.1.6 Constants

Con tant are numeric quantitie in a proce model who e value do not change. For example,

area of a heat exchanger and molecular weight of a component are con tant . In the Flow heet

Simulation program, con tant are entered are tored in logical group called ‘con tant property’.

Con tant propertie can have name and de cription . For example, area of the heat exchanger in the

model Alkyl are tored together in a con tant property called ‘ calar_area ’, with a de cription of

‘Heat Exchanger Area ’. The program’ u er manual de cribe thi in detail, and the con tant u ed for

the alkylation proce are given in Appendix C.8.

5.1.7 Enthalpy Tables

An enthalpy table in Flow heet Simulation program can tore enthalpy coefficient of the

chemical component in the proce model. The e table are u ed in On-Line Optimization and Pinch

Analy i program . The enthalpy formula u ed here i :

2 3 4 5 Enthalpy = A0 + A1*T + A2*T + A3*T + A4*T + A5*T

where, A0 to A5 are enthalpy coefficient and T i the temperature. Any con i tent et of unit can be

u ed here. The program’ u er manual de cribe entering thi information in detail. Enthalpy

coefficient value for ga and liquid pha e entered into Flow heet Simulation program are given in

Appendice C.6 and C.7.

93 5.1.8 Flowsheet Simulation Summary

The alkylation proce model developed u ing the Flow heet Simulation program i

ummarized in Table 5.1. The degree of the freedom in the model i 55.

Table 5.1. Summary of the Alkylation Proce model

Feature Quantity

Proce Unit 76 Proce Stream 110 Equality Con traint 1579 Inequality Con traint 50 Mea ured Variable 125 Unmea ured Variable 1509 Parameter 64

5.2 On-Line Optimization

A di cu ed in Chapter 2 and 3, On-Line Optimization u e the plant model developed in

Flow heet Simulation to calculate optimal etpoint for the di tributed control y tem. Thi involve rectifying gro error of plant data ampled from di tributed control y tem u ing combined gro error detection and data reconciliation, e timating proce parameter and reconciling plant data u ing

imultaneou data reconciliation and parameter e timation, and optimizing the operating etpoint u ing the updating proce and economic model . The e three tep were hown in Figure 1.2.

5.2.1 Gross Error Detection and Data econciliation

Combined gro error detection and data reconciliation i the fir t tep in conducting on-line optimization. A mentioned in Chapter 2 and 3, On-Line Optimization olve thi tep by creating a non-linear optimization problem, where the proce model erve a the et of con traint , and the objective function i one of the available method pecified by the u er. The program olve the

94 optimization problem by u ing GAMS. The program’ u er manual de cribe thi procedure in detail.

In thi tep the data i reconciled and gro error are detected and removed. Their value are replaced by reconciled value , and thi give a et of data with only random error for u e in data reconciliation

and parameter e timation.

For the alkylation proce model, Robu t Function method wa elected a the objective

function and CONOPT2 wa et a the default olver for GAMS. The program gave an optimal

olution of 78.8 after 1,192 iteration for the operation point #1 of the ix teady ate operation point .

The other had comparable value . The reconciled data and the optimal etpoint for the mea ured

variable are given in the Table 5.2 for operation point #1. For a confidence level of 95% the critical

value i calculated a 3.53, i.e. if the tandard mea urement error (∈i = |yi -xi |/σi) i greater than 3.53 a

gro error i declared. U ing thi te t criterion 31 mea urement are identified a having gro error

a hown in Table 5.2. Appendix D.1 and D.2 document the reconciled value from data validation for

mea ured and unmea ured variable for the alkylation proce model. Appendix H.1 document the

complete GAMS output for the Data Validation tep.

Table 5.2. Mea ured Variable for operation point #1

Variabl Nam Plant Data Optimal S t Point R concil d Data R concil d Data Gross From Param t r From Data Validation Error

Estimation |yi -xi |/σi FAC02 0.1125 0.1600 0.1600 0.1600 4.2235 FAC12 0.1259 0.1600 0.1600 0.1600 FAC23 0.1253 0.1600 0.1600 0.1600 FAC34 0.1389 0.1600 0.1600 0.1600 FAC45 0.1040 0.1600 0.1600 0.1600 5.3846 FC308 2.1990 3.1197 3.0606 3.1032 4.1120 FC316 0.6581 1.7518 1.8000 1.8000 17.3515 FC320 0.1468 0.0459 0.1490 0.1497 FC322 0.4427 1.5684 1.5644 1.5619 25.2812 FC328 0.0942 0.0545 0.0547 0.0535 FC329 0.7724 0.7635 0.7655 0.7491

95 FC403 3.8766 2.3677 2.3223 2.2834 4.1097 FC407 1.1070 0.9348 0.9205 0.9145 FC412 0.0324 0.0420 0.0421 0.0418 FC417 0.2799 0.1439 0.2754 0.2748 FHC01 1.0155 0.8980 0.8569 0.8513 FHC32 1.8596 1.9569 1.8455 1.8690 FSC402 0.5223 0.4587 0.4916 0.4943 FSC405 0.3344 0.3148 0.3452 0.3463 FSC411 2.7287 1.3267 1.3499 1.3525 5.0436 FSC413 0.1445 0.1439 0.1464 0.1480 FstmE612 0.1425 0.0888 0.0889 0.0889 3.7607 PC302 102.3615 101.0000 102.8996 102.1638 PC310 264.5339 260.0000 264.1366 264.5758 PC601 618.4222 625.0000 625.0000 625.0000 PC603 1635.3720 1691.3731 1694.5308 1694.5255 QHC07 2.2343 1.9469 1.7391 1.8842 QHC11 1.8496 1.7391 1.7761 1.7391 QHC14 2.0061 1.7391 1.9149 1.7391 QHC16 2.0344 1.7391 1.7391 1.7391 QHC34 0.9205 1.1464 0.9354 0.9505 QHC38 0.5705 0.5352 0.5761 0.5791 QHC41 0.8344 0.8699 0.8465 0.8489 QHC45 0.8392 0.8518 0.8515 0.8720 TAC09 284.1800 280.0000 281.6669 282.0713 TAC12 284.5783 280.0000 281.6669 282.0713 TAC23 284.7406 280.0000 280.0000 281.4793 TAC31 282.9628 280.0000 281.1345 281.4267 TAC34 284.9250 280.0000 281.1345 281.4267 TAC42 287.1733 280.0000 283.3783 284.2594 TAC45 284.9372 280.0000 283.3783 284.2594 TC303 264.0644 280.6862 280.6987 281.4996 TC306 341.9278 349.4271 349.7839 349.8897 TC307 315.5806 328.4033 329.3628 328.9195 TC308 315.8461 328.4033 329.3628 328.9195 TC315 308.2378 307.3612 308.2378 308.6556 TC316 344.0172 345.2250 345.1410 345.1528 TC317 363.4556 359.0690 359.0000 359.0000 TC321 301.0450 300.0000 301.0948 301.1193 TC324 365.0522 359.0690 359.0000 359.0000 TC325 319.6617 322.5997 322.6861 322.6859 TC404 318.6700 305.2492 306.2880 307.0948 TC405 426.1200 410.0000 411.1033 410.0000 TC407 302.9500 303.5753 302.9500 302.9500 TC408 426.1200 405.0000 405.0000 405.0000 TC410 356.2411 363.3260 363.3927 363.0366 TC414 324.4561 337.0859 336.7854 336.7774

96 TC418 305.9128 307.1433 307.4867 308.4341 TC419 301.6367 304.5245 305.1181 305.9727 THC32 263.4350 261.9077 256.5978 256.4405 TSC402 314.6989 322.8886 324.6526 324.6567 TSC403 337.5928 336.6046 335.7424 335.7490 TSC405 301.2561 300.0000 301.2561 301.2561 TSC408 319.2900 318.7440 317.9076 317.9181 TSC413 298.6917 300.0000 300.0000 300.0000 x11AC12 0.9670 0.9687 0.9718 0.9695 x11AC23 0.9380 0.9424 0.9451 0.9432 x11AC34 0.9220 0.9162 0.9162 0.9169 x11AC45 0.9060 0.8900 0.8900 0.8906 x1C316 0.1408 0.1166 0.1184 0.1178 x1C325 0.9720 1.0000 1.0000 1.0000 x1C417 0.0372 0.0200 0.0253 0.0255 x1HC32 0.0258 0.0228 0.0235 0.0233 x1SC402 0.0143 0.0063 0.0142 0.0142 x1SC403 0.0001 0.0000 0.0000 0.0000 x1SC408 0.0475 0.0200 0.0477 0.0474 x2SC402 0.0136 0.0000 0.0084 0.0084 3.7929 x2SC403 0.0119 0.0000 0.0119 0.0119 x2SC408 0.0221 0.0000 0.0002 0.0002 9.9048 x3C316 0.7790 0.7851 0.7880 0.7886 x3C325 0.0017 0.0000 0.0000 0.0000 10.0000 x3C417 0.8699 0.9738 0.8137 0.8141 x3HC32 0.8187 0.7604 0.7708 0.7708 x3SC402 0.2910 0.3216 0.2959 0.2973 x3SC403 0.0103 0.0234 0.0212 0.0212 10.5665 x3SC408 0.8761 0.9739 0.9435 0.9436 x4C316 0.0580 0.0894 0.0795 0.0796 3.7155 x4C417 0.0609 0.0061 0.0514 0.0509 x4HC32 0.1054 0.1401 0.1255 0.1257 x4SC402 0.5934 0.5280 0.5601 0.5587 x4SC403 0.8540 0.7666 0.7940 0.7936 x4SC408 0.0331 0.0061 0.0086 0.0088 7.3475 x5C316 0.0020 0.0056 0.0060 0.0060 19.8000 x5C417 0.0009 0.0000 0.0295 0.0295 286.2300 x5HC32 0.0096 0.0288 0.0306 0.0306 22.0134 x5SC402 0.0503 0.0686 0.0516 0.0516 x5SC403 0.0758 0.1000 0.0734 0.0736 x5SC408 0.0001 0.0000 0.0000 0.0000 x6SC402 0.0167 0.0686 0.0666 0.0666 29.8204 x6SC403 0.0250 0.1000 0.0949 0.0950 27.9946 x6SC408 0.0001 0.0000 0.0000 0.0000 x7HC32 0.0197 0.0479 0.0496 0.0497 15.2312 x7SC402 0.0022 0.0069 0.0032 0.0032 4.3956

97 x7SC403 0.0046 0.0100 0.0046 0.0046 x7SC408 0.0022 0.0000 0.0000 0.0000 10.0000 xx1C322 0.0027 0.1133 0.1167 0.1167 428.5338 xx1C414 0.0330 0.0762 0.0798 0.0800 14.2498 xx1HC01 0.1088 0.0976 0.1111 0.1099 xx2HC01 0.4525 0.1296 0.1290 0.1291 7.1481 xx3C317 0.9430 0.7900 0.7930 0.7930 xx3C322 0.9430 0.7900 0.7930 0.7930 xx3C407 0.0003 0.0000 0.0000 0.0000 7.4194 xx3C412 0.0027 0.0021 0.0027 0.0027 xx3C414 0.8772 0.8081 0.8196 0.8196 xx3HC01 0.3558 0.0125 0.0125 0.0125 9.6498 xx4C317 0.0671 0.0899 0.0800 0.0800 xx4C322 0.0671 0.0899 0.0800 0.0800 xx4C407 0.1124 0.0851 0.0850 0.0853 xx4C412 0.9004 0.8771 0.8668 0.8549 xx4C414 0.0860 0.1067 0.0916 0.0914 xx4HC01 0.1085 0.1108 0.1064 0.1067 xx5C407 0.0803 0.1512 0.1514 0.1506 8.7555 xx5C412 0.0022 0.0568 0.0581 0.0581 255.6751 xx5C414 0.0021 0.0010 0.0011 0.0011 4.8325 xx7C414 0.0015 0.0080 0.0080 0.0080 44.4218

5.2.2 Parameter Estimation and Data econciliation

In thi tep the data i reconciled and parameter e timate are updated by olving the nonlinear programming problem u ing the mea ured variable value from the previou tep. The program gave

an optimal olution of 113.8 after 1,490 iteration for the operation point #1 of the ix teady tate ,

and the e re ult are hown in Table 5.2 for the mea ured variable . The performance for the other five

teady tate operation point were imilar. The updated value of the parameter for the fir t teady

tate olution are li ted in Table 5.3. The value of 36 of the 64 parameter remained the ame wherea

the adju tment for the re t during parameter e timation are minimal. The e value along with the

error free, reconciled mea ured variable repre ent the current pecification of the proce plant,

which can be u ed to calculate the optimal operating etpoint . Appendix H.2 document the complete

GAMS output for the Parameter E timation tep.

98 Table 5.3. Plant Parameter for operation point #1

Plant Param t r Initial Valu Estimat d Valu d ltaPE634 70.0000 70.0000 d ltaPE640 20.0000 20.0000 FE601 0.5000 0.5000 FE603 1.0000 1.0000 FE609A 0.5000 0.5000 FE610 1.0000 0.9340 FE611 0.5000 0.5218 FE616 0.5000 0.5000 FE617 1.0000 1.0000 FE621A 0.7220 0.7397 FE621B 1.0000 1.0000 FE626 0.5000 0.5000 FE627A 0.5000 0.5016 FE627B 0.5000 0.5257 FE628 0.5000 0.5000 FE629 0.5000 0.5000 FE634 1.0000 1.0000 FE640 0.5000 0.7569 FE641 0.5000 0.5000 hstmE602 2145.0000 2145.0000 hstmE612 2145.0000 2145.0000 hstmE695 1920.0000 1920.0000 hstmE696 2145.0000 2145.0000 PC606A 900.0000 900.0000 PC606C 890.0000 890.0000 PC606D 900.0000 900.0000 PE633 145.0000 145.0000 qC601 1.0000 1.0000 qC603 1.0000 0.9904 qC606A 0.5000 0.5000 RC601 9.1410 9.2210 RC603 14.0000 14.0000 sfC631 0.9770 0.9774 sfC632 0.9820 0.9815 sfC633 0.9910 0.9913 sfC634 0.9900 0.9893 Tcwin 290.0000 290.0000 UE601 0.0080 0.0080 UE602 0.0160 0.0160 UE603 0.0250 0.0251 UE605 0.0450 0.0498 UE609A 0.0400 0.0400 UE610 0.0830 0.0898

99 UE611 0.0990 0.1000 UE612 0.0130 0.0130 UE613 0.0200 0.0256 UE616 0.0100 0.0100 UE617 0.0520 0.0526 UE621A 0.1140 0.1140 UE621B 0.0780 0.0759 UE626 0.0100 0.0100 UE627A 0.0100 0.0100 UE627B 0.0100 0.0100 UE628 0.0100 0.0144 UE629 0.0100 0.0100 UE633 0.0160 0.0177 UE634 0.0210 0.0206 UE640 0.0100 0.0100 UE641 0.0840 0.0874 UE695A 0.0330 0.0329 UE695B 0.0390 0.0388 UE696A 0.0120 0.0122 UE696B 0.0100 0.0100 UE6XX 0.0310 0.0286

5.2.3 Economic Optimization

Maximizing profit wa u ed a the objective of economic optimization of the alkylation proce model. The alkylation unit co t and price for ummer and winter condition are a hown in

Table 5.4. The objective function wa ba ed on the winter data ince the plant data available wa

mea ured in winter.

The economic model wa developed a follow :

Profit = Sale - Co t - Utilitie

Sale = Alkylate (C3, C4 and C4 Raffinate) produced * Price of alkylate

Co t = Σ Input * Co t

Utilitie = Σ Input * Utility Co t

100 where the Input for Co t include olefin (propylene and butylene), C4’ from the reformer (feed to the

Saturate Dei obutanizer column), i obutane and ulfuric acid and the Input for the Utilitie include

team, water and electricity.

Thi economic model wa u ed with operation point #1 in Appendix I, and the program gave

an optimal olution after 63 iteration . The profit for the proce wa calculated to be $29.11/min,

which i an increa e of 144% over the operating condition (#1). The profit from the current operating

condition wa evaluated u ing the reconciled data prior to economic optimization. Thi improvement

in the profit i cau ed by 8.5% reduction in co t and 2.2% increa e in ale . The economically

optimum olution had 5.5% more olefin charge, almo t 100% reduction in i obutane purcha e co t

(becau e of reduced i obutane flow rate), 7.2% reduction in aturate feed to the Saturate

Dei obutanizer column and 2.2% increa e in the alkylate. The alkylate quality did not change at the

economically optimal operation. Energy aving attained through reduced team u age in the

di tillation column total an average of 9.4x109 BTU/yr. Thi amount corre pond to 0.16% reduction

in the utility co t and re ult in a aving of $5600 /yr.

The re ult for all of the ix ca e are hown in Table 5.5 where the increa e wa from 25% to

216%. Thi wide range of increa e in the optimal profit for ix different operation point i ob erved becau e of even wider range exi ting in the plant data. The flow mea urement differ a much a

300% and ma fraction mea urement a much a 4000%, between operation point . Appendice

D.1 and D.2 document the optimal value for mea ured and unmea ured variable for the alkylation

model. Appendix H.3 document the complete GAMS output for thi tep.

Collectively, the e re ult how that by applying on-line optimization to the alkylation proce

with reconciled data and e timated parameter , the profit of the plant can be improved ignificantly.

101

Table 5.4. Alkylation Plant Raw Material/Utility Co t and Product Price

Feed and Product Stream Co t and Price ($/bbl) Number Summer Winter

Feeds Propylene HC01 11.79 10.44 Butylene HC01 18.00 16.56 I o-butane SC414 16.88 17.39

Products N-butane SC405, C413 13.29 12.71

C3 Alkylate C407 24.49 22.30

C4 Alkylate C407 26.32 24.06

C4 Raffinate Alkylate C407 26.34 24.19

Cataly t and Utilitie Co t

H2SO4 (Stream AC02) $110/Ton Electricity $0.04/KWH 50# Steam $2.50/M-Lb 250# Steam $3.60/M-Lb 600# Steam $4.40/M-Lb

102

Table 5.5. Calculated Profit after Data Validation (D.V.), Parameter E timation (P.E.) and Economic

Optimization (E.O.) Step for ix Different Operation Point (Steady State )

Operation point D.V. P.E. E.O % Increase

#1 11.9 12.1 29.1 144

#2 7.4 7.4 21.4 189

#3 21.4 22.1 26.9 26

#4 7.0 7.0 22.1 216

#5 10.1 23.3 26.3 160

#6 22.0 23.6 27.6 25

Average % increa e 127

5.3 Heat Exchanger Network Optimization

The Heat Exchanger Network Optimization (THEN) program integrate network of heat

exchanger , boiler and conden er . It u e pinch analy i a the ba i for de igning an optimum

network. The Pinch Analy i program of Advanced Proce Analy i Sy tem i u ed to perform off-

line analy i of proce e to achieve better energy integration. Thi ection de cribe the u e of the program for the alkylation proce .

Alkylation proce i very energy inten ive. The alkylation reaction occurring in the contactor are exothermic, and the heat generated i removed by effluent refrigeration. The proce require proper control of temperature, which i done by feed-effluent heat exchanging and al o by external utilitie . Al o, energy i required in the eparation unit of the proce . The alkylation proce

103 model ha 28 heat exchanger , plu four pair of contactor . The heat exchange within the contactor u ing the cold refrigerant conden ate through the tube bundle i not included in the pinch analy i

ince any new arrangement for the contactor will be impractical.

A di cu ed in Chapter 3, the fir t tep in u ing THEN i the identification of tream important for heat integration. The e tream are the input and output tream to heat exchanging unit . Appendix E contain the li t of tream from the model u ed for heat integration.

The econd tep i retrieving the required information for the elected tream, from the optimization re ult . The data retrieved are temperature and flowrate a hown in the u er’ manual,

Appendix J. Energy calculation can be done either u ing con tant heat capacitie or pecifying enthalpy coefficient for variable heat capacitie . To obtain average value for con tant heat capacitie , enthalpy of the tream hould be u ed and thi calculation i executed in the program a illu trated in the u er’ manual. For variable heat capacitie , Appendice C.6 and C.7 contain the coefficient of the equation u ed to evaluate the enthalpy for the component in the proce for ga and liquid pha e .

The temperature-dependent enthalpy equation u ed i a follow :

2 3 4 5 H = A0 + A1 T + A2 T + A3 T + A4 T + A5 T MJ/min (5.1)

From the enthalpy coefficient and tream compo ition , average enthalpy coefficient for tream are calculated a :

Aavg = Σ AI xI (5.2) where, AI and xI are the enthalpy coefficient and mole fraction of component I.

The third tep i cla ifying the tream a hot or cold tream . A mentioned in Chapter 3, a hot tream i a tream that need to be cooled, and a cold tream i a tream that need to be heated.

The e are de ignated in the program a hown in the u er’ manual. Finally, the minimum approach temperature for the network i cho en to be 13 K (See u er’ manual). Thi i to en ure ufficient

104 driving force between the tream in the network. The program u e thi information to conduct pinch

analy i for the alkylation proce by u ing the FORTRAN program THEN.

One of the re ult of the program i the Grand Compo ite Curve (GCC) hown in Figure 5.1.

Thi diagram how the temperature a a function of enthalpy for a compo ite of the tream . The

region of the GCC with a po itive lope how the proce egment that require energy, wherea the

region that ha negative lope how the egment that reject heat. The end point of the curve give

the minimum value of external heating and cooling required by the proce .

According to the Pinch Analy i the alkylation proce require a minimum of 1742 MJ/min

external heat and 4043 MJ/min of external cooling. From the data validation re ult , the current

external utility requirement are 1907 MJ/min of heat and 4300 MJ/min of cooling. The economic

optimization decrea e the heating requirement by 1% to 1888 MJ/min wherea the initial pinch

analy i reduce it another 7.7%. The cooling requirement can be reduced a much a 7.4% by u ing pinch analy i from 4367 MJ/min after economic optimization. Thi i becau e the economic

optimization re ult in a 1.6% higher cooling requirement than the current value of 4300 MJ/min.

Figure 5.1 Grand Compo ite Curve for Alkylation Proce

105 Pinch Analy i program al o ha the ability to de ign a Maximum Energy Recovery (MER)

network for the proce under con ideration. The network grid diagram that make u e of the external

utilitie calculated in the GCC can be een in Figure 5.2. The network found by pinch analy i con i t

of 16 heat exchanger , 4 heater and 15 cooler , wherea the proce ha only 6 heat exchanger , 4

heater and 12 cooler . Thi ugge t that the improvement in the energy requirement are achieved by

the e additional heat exchanger .

Heat integration above the pinch involve tream uch a SC406 (to Saturate Dei obutanizer

reboiler), C315 (charge to Depropanizer column) and C410 ( ide tream to inter-reboiler of

Dei obutanizer column) which are heated up by tream C405, C406 (Dei obutanizer bottom ), C317

(Depropanizer bottom ) and C412 ( ide tream from De iobutanizer). Thi integration eliminate ome

of the heat exchanger exi ting in the current plant configuration (e.g. E-616). However, the

configuration from the analy i may re ult in operational difficultie becau e of a more inten e

interaction between input and output tream of the three di tillation column . Moreover, the e three

di tillation column are placed acro the pinch, which i not an appropriate placement of di tillation

column for energy integration.

To integrate the column with the remainder of the proce , one can remove the column from

the proce and then try to u e a much energy a po ible from the proce for the energy

requirement of the column by pre ure- hift (Dougla , 1988). A pre ure hift applied to column 5C-

601 (a decrea e in the operation temperature by 7 K) can reduce the heating and cooling requirement by 550 MJ/min. Pre ure hift re ulting in 25 K and 9 K decrea e in operation temperature for 5C-

603 and 5C-601 can reduce the eparation energy requirement by 650 MJ/min (Figure 5.3). The e change hould be con idered if it i fea ible with the other operating condition in the plant.

106 From the Pinch Analy i , three loop and one path in the heat exchanger network can be located. The e provide additional degree of freedom for further optimization of the y tem by eliminating ome of the exchanger within the loop or on the path.

In ummary, Pinch Analy i provided an exten ive in ight for the optimization of the energy con umption in the alkylation plant and howca ed the benefit of heat integration for the proce .

Figure 5.2. Network Grid Diagram for Alkylation Proce

107

400 400

350 350

C601 C601

300 300

280 0 400 800 1200 1600 2000 280 MJ/min 0 400 800 1200 1600 2000 MJ/min

Figure 5.3. Integrating column (5C-601 and 5C-603) with the proce : Pre ure hift for column 5C-601 only (left), for both column (right).

108 5.4 Pollution Assessment Pollution A e ment program i u ed to a e the pollutant generated in the proce and determine the location where they are generated. Thi information i u ed to modify the proce for wa te minimization. Al o, a pollution index i evaluated for compari on of different proce e . In the fir t tep of the program, input and output tream of the proce are pecified a hown in the u er’ manual, Appendix J.

The alkylation proce ha 10 input and output tream a li ted in Table 5.6. The e tream are

elected from Flow heet Simulation. The output tream are cla ified a product and non-product. For example, pent acid (AC45) i a non-product wherea alkylate (C407) i a product tream. The component pre ent in each of the e tream are pecified and the flow rate and compo ition of

tream are obtained from the re ult of On-Line Optimization program. The program’ u er manual de cribe the e tep in detail.

Table 5.6. Input and Output Stream in Alkylation Proce . Stream De cription Type Pollution Index

AC02 Fre h Acid Feed Input 0.808 HC01 Olefin Feed Input 1.622 SC414 Make-up I obutane Input 1.611 SC401 Sat-Dei obutanizer Feed Input 1.789 AC45 Spent Acid Non-Product 1.034 C320 To LPG Storage Product 0 C328 To Fuel Ga Product 0 C407 To Alkylate Storage Product 0 C413 To N-butane Storage Product 0 SC405 To N-butane Storage Product 0

109 A di cu ed in Chapter 3, pollution impact i calculated u ing pecific environmental impact potential (SEIP) of the component in the tream . SEIP value of the component in alkylation proce are li ted in Appendix G. Relative weighting factor for the nine categorie of impact were all

a umed to be one in the ab ence of actual value . U ing the SEIP value and relative weighting

factor the program calculate pollution indice for each input, product and non-product tream in the proce , caling the effect of the tream to the environment. The e value are u ed to calculate the ix pollution indice for the proce , which are li ted in Table 5.7, before (BEO) and after (AEO) the

economic optimization of the proce . Negative value mean that the input tream are actually more

harmful to the environment than the non-product if they are not proce ed through the alkylation proce .

Table 5.7. Pollution A e ment Value for Alkylation Proce before (BEO) and after (AEO)

the economic optimization.

Index Type Value (BEO) (AEO) Total rate of impact generation -4.9120 -4.7966 impact/time Specific impact generation -3.2860 -3.4584 impact/product Pollution generation per unit product -0.9777 -0.9742 ma of pollutant/ma of product Total rate of impact emi ion 1.0325 1.0337 impact/time Specific impact emi ion 0.6897 0.7453 impact/product Pollutant emi ion per unit product 0.1069 0.1154 ma of pollutant/ma of product

Pollution A e ment re ult how that the economic improvement that i achieved by the economic optimization doe n’t come with a reduced environmental impact. The plant operating at the optimal et point emit more pollutant ince the rate of impact generation i reduced, although

pecific component’ con umption might be le (e.g. ulfuric acid con umption i reduced by 2.2%).

110 5.5 Summary

Advanced Proce Analy i Sy tem wa applied to the Motiva alkylation plant. The proce model wa developed u ing Flow heet Simulation program. On-Line Optimization wa u ed to generate optimal et point for the di tributed control y tem. Pinch Analy i program wa u ed for energy integration and to determine the minimum heating and cooling requirement . A pollution a e ment of the plant wa made to determine the ource of the pollution generation.

111

CHAPTE 6 CONCLUSIONS

Alkl tion process is one of the most import nt refinery processes for producing convention l g soline. This project demonstr tes process control nd modific tion improvements to this process with Adv nced Process An lysis System th t employs process flowsheeting, on-line optimiz tion, pinch n lysis nd pollution ssessment.

Also, the chemic l re ction n lysis p rt of the System w s used to determine the best

lkyl tion re ction kinetics.

Using the flowsheeting c p bility of the Adv nced Process An lysis System simul tion of the lkyl tion process w s developed th t consist of 76 process units, 110 process stre ms, 1579 equ lity nd 50 inequ lity constr ints with 1634 v ri bles. The simul tion w s v lid ted using pl nt d t nd d t reconcili tion to show th t the simul tion predicted the perform nce of the pl nt within the ccur cy of the d t .

The n lysis of the pl nt d t resulted in detecting six ste dy st te oper tion points. For e ch oper tion point gross errors were detected, d t were reconciled, p r meters were upd ted nd economic lly optimum setpoints re determined for the distributed control system.

The economic optimiz tion of the process for six oper tion points resulted in

25.4% to 215.4% incre se in the profit. As n ex mple; the profit for the process w s c lcul ted to be $29.11/min, which is n incre se of 144.6% over the oper ting condition (#1). This improvement in the profit is c used by 8.5% reduction in costs nd

2.2% incre se in s les. The economic lly optimum solution results in 5.5% more olefin ch rge, lmost 100% reduction in isobut ne purch se cost, 7.2% reduction in s tur te feed to the S tur te Deisobut nizer column nd 2.2% incre se in the lkyl te. The

112

lkyl te qu lity did not ch nge t the economic lly optim l oper tion. Energy s vings tot l n ver ge of 9.4x109 BTU/yr. This mount corresponds to 0.16% reduction in the utility cost nd results in s ving of $5600 /yr. Another result obt ined from the economic optimiz tion of the lkyl tion process is 2.2% reduction in the sulfuric cid consumption.

According to the Pinch An lysis the lkyl tion process requires minimum of

1742 MJ/min extern l he t nd 4043 MJ/min of extern l cooling. From the d t v lid tion results, the current extern l utility requirements re 1907 MJ/min of he t nd

4300 MJ/min of cooling. The economic optimiz tion decre ses the he ting requirement by 1% to 1888 MJ/min where s the initi l pinch n lysis reduces it nother 7.7% ( tot l of 67x109 BTU/yr). The cooling requirement c n be reduced s much s 6% ( tot l of 106x109 BTU/yr). by using pinch n lysis . A further reduction in the energy requirements c n be chieved by n ppropri te pressure shift pplied to distill tion columns ccounting m ximum reduction of 650 MJ/min (268x109 BTU/yr).

Pollution ssessment of the lkyl tion pl nt reve led the extent nd loc tion of the pollut nt emissions of the process. It h s lso shown th t the economic lly optim l solution c n result in higher over ll pollution levels even if the consumption of the sulfuric cid is reduced.

113

REFERENCES

Albright, L. F., 1990a, Modern Alkylate - 1:Alkylation will be key process in reformulated gasoline era, Oil and Gas Journal, 88, no. 46, 79-92

Baker, J. R., R. J. Hossli, M. M. Zanoni and P. P. Radeciki, 1995, A Demonstration Version of The Clean Process Advisory System, Paper No. 8a, AIChE Summer National Meeting, Boston

Ballinger, G. F., R. Banares-Alcantara, D. Costello and E. S. Fraga, 1994, epee: A Process Engineering Software Environment, Computers Chem. Engg., Vol. 18, Supp. P. S283-S287

Bernado N., M. Mazzotti, G. Storti, and M. Morbodelli, 1998 Multicomponent Distillation Design through Equilibrium Theory, Ind. Eng. Chem. Res., 37, 2250-2270

Brook, A., D. Kendrick, A. Meeraus, 1998, GAMS- A User’s Guide, Release 2.25, The Scientific Press

Cabezas, H., J. C. Bare and S. K. Mallick, 1997, Pollution Prevention with Chemical Process Simulators: The Generalized Waste Reduction Algorithm, Computers Chem. Engng., Vol. 21, Supp. p. S305-S310

Chen, X., 1998, The Optimal Implementation of On-Line Optimization for Chemical and Refinery Processes, Doctoral Dissertation, Louisiana State University, Baton Rouge, LA

Corma, A. and A. Martinez, 1993, Chemistry, Catalysts and Processes for Isoparaffin - Olefin Alkyalation: Actual situation and Future trends, Catalysis Reviews: Science and Engineering, 35, no. 4, 485-523

Cupit, C. R., J. E. Gwyn and E. C. Jernigan, 1961, Catalytic Alkylation, Petro/Chem Engineer, 33, no. 47

Cupit, C. R., J. E. Gwyn and E. C. Jernigan, 1962, Catalytic Alkylation, Petro/Chem Engineer, 34, no. 49

Douglas, M. James, 1988, Conceptual Design of Chemical Processes, McGraw-Hill, Inc.

Gary, H. James and H. E. Glenn, 1984, Petroleum Refining, Second Edition, Marcel Dekker, Inc.

Graves, D. C., 1999, Acid consumption and acid runaway on a sulfuric acid alkylation unit, Alkylation Seminar in Scottsdale, Arizona.

114

Gruse, W. A. and D. R. Stevens, 1960, Chemical Technology of Petroleum, Third Edition, McGraw-Hill, New York

Gundersen, T. and Naess, L., 1988, The Synthesis Of Cost Optimal Heat Exchanger Networks, Computers Chem. Engng., Vol. 12, No. 6, p. 503-530 Hegde, S. C., 1999, Simulations for Pollution Prevention, Master’s Thesis, Lamar University, Beaumont, TX

Heijungs, R. (Final Ed.); Guinee, J. B., Huppes, G., Lankreijer, R. M., Udo de Haes, H. A. and Sleeswijk, A. W., 1992, Environmental Life Cycle Assessment of Products Guide - October 1992, Centre of Environmental Science, Leiden

Hilaly, A. K. and Sikdar, S. K., 1994, Pollution Balance: A New Methodology for Minimizing Waste Production in Manufacturing Processes, J. Air and Waste Manage. Assoc., Vol. 44, p. 1303-1308

Hohmann, E. C., 1971, Optimum Networks for Heat Exchange, Ph.D. Dissertation, University of Southern California.

Kister, Distillation Design, McGraw-Hill Publishers

Langley, J. R., 1969, The Kinetics of Alkylation of Isobutane with Propylene Using Sulfuric Acid Catalyst, Doctoral Dissertation, Louisiana State University, Baton Rouge, LA

Langley, J. R. and R. W. Pike, 1972, The Kinetics of Alkylation of Isobutane with Propylene, AIChE Journal, 18, no. 4, 698-705

Lee, K.F., A. H. Masso and D. F. Rudd, 1970, Branch and Bound Synthesis of Integrated Process Design, Ind. Engng. Chem. Fundam., Vol. 9, p. 48-58

Linnhoff, B., D. R. Mason and I. Wardle, 1979, Understanding Heat Exchanger Networks, Computers Chem. Engng., Vol. 3, p. 295-302

Masso, A. H. and D. F. Rudd, 1969, The Synthesis of System Designs. II. Heuristic Structuring, AIChE Journal, Vol. 15, p. 10-17

McBride, B. J., S. Gordon, M. A. Reno, 1993, Coefficients for Calculating Thermodynamic and Transport Properties of Individual Species, NASA Technical Memorandum 4513

Perry, R. H., Green, D., Malony, J. O. (editors), 1984, Perry’s Chemical Engineers’ Handbook, Sixth Edition, McGraw-Hill International Publishers

Petrol.Refiner., 1952, 31 (9), 156-164

115

Rosenwald, R. H., 1978, Alkylation, Kirk-Othmer Encyclopedia of Chemical Technology, Third Edition, John Wiley & Sons, New York

Saleh, J. M., J. R. Hopper and R. E. Walker, 1995, Three-Phase, Catalytic Gas-Liquid Reactors: An Interactive Simulator, Paper No. 73d, AIChE Spring National Meeting, Houston.

Shaney, Agnes, 1995, Cleaner by Design, Chemical Engineering, Vol. 102, No. 7, p. 32.

Shenoy, U. V., 1995, Heat Exchanger Network Synthesis: Process Optimization by Energy and Resource Analysis, Gulf Publishing Co., Houston, Texas

Smith, B. D., 1963, Design of Equilibrium Stage Processes, McGraw-Hill Book Company, Inc., New York.

Telang, K., 1998, Advanced Process Analysis System, Master’s Thesis, Louisiana State University, Baton Rouge, LA

Templeton, P. C. and B. H. King, 1956, paper presented before the Western Petroleum Refiners Association, June 21-22

Thomas, C. L., 1970, Catalytic Processes and Proven Catalysts (McGraw Hill Book Company, New York), pp. 87-96

Van Reeuwijk, P. A., R. Zwart and M. E. J. Dreissen, 1994, CAPE in the Development of an Energy Efficient Chemical Process, Computers Chem. Engg., Vol. 18, Suppl. P. S161-S164

Vichailak, M., 1995, Pollution Prevention by Process Modification, D. E. Field Study, Lamar University, Beaumont, TX

Warren L. McCabe, Julian C. Smith, Peter Harriott, 1993, Unit Operations of Chemical Engineering, Fifth Edition, McGraw-Hill, Inc.

Winter, Peter, 1992, Computer Aided Process Engineering: The Evolution Continues, Chem. Engng. Prog., Vol. 88, No. 2, p. 76-83

Yaws, C. L., 1999, Chemical Properties Handbook, McGraw-Hill Handbooks, New York

Yongkeat, S, 1996, The sulfuric Alkylation of isobutane with light olefins, Master’s Dissertation, Lamar University, Beaumont, TX

Young, D. M., 1999, Personal Communication

116

117